naviserver-tracker — SourceForge Tracker Notifications

[ naviserver-Feature Requests-1171843 ] nsd -c: start a REPL

[SourceForge.net <no...@so...>]

Feature Requests item #1171843, was opened at 2005-03-28 13:43
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1171843&group_id=130646

Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: nsd -c: start a REPL

Initial Comment:

Here's a patch which adds a new startup mode to the nsd
binary: -c command mode.  When selected, after the
server has started rather than eternaly wait for
signals it runs a Tcl shell and waits for commands.

The nice thing about this is that all the server
commands are available, which is not the case if you
simply load libnsd.so.

The main goal is testing.  I've done some cleanup and
you can now type 'make test' in the top level
directory.  'make runtest' will drop you into an
interactive shell.

It could do with some improvements: 'exit' should
probably trigger an orderly shutdown, and signals
aren't handled.

Unfortunately ns_geturl is spectacularly useless,
simply fetching the bytes of a file directly from the
file system.  Is this useful?  I would have expected it
to pass through registered procs etc.

I'm thinking maybe the way to go is an nstest module
which adds some extra commands to dummy up a connection
and run the registered proc.  It would be possible I
guess to also create nstestdb to create a dummy db
driver, and so on...


Feedback appreciated.



----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-28 16:38

Message:
Logged In: YES 
user_id=184124

I usually use nscp & socat combination to test the server,
in this mode it runs as it should and all commands are
tested using the same environment as it will be in the
production.
In your mode, server is used like shared library, to test
commands interactively only, you still need to run the
server to test background procs/callbacks.
Is it possible to run server in foreground (-f) and spawn
Tcl shell at the same time. In this case, you have the
server and nscp-like shell at the same time.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1171843&group_id=130646

[ naviserver-Feature Requests-1171843 ] nsd -c: start a REPL

[SourceForge.net <no...@so...>]

Feature Requests item #1171843, was opened at 2005-03-28 06:43
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1171843&group_id=130646

Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: nsd -c: start a REPL

Initial Comment:

Here's a patch which adds a new startup mode to the nsd
binary: -c command mode.  When selected, after the
server has started rather than eternaly wait for
signals it runs a Tcl shell and waits for commands.

The nice thing about this is that all the server
commands are available, which is not the case if you
simply load libnsd.so.

The main goal is testing.  I've done some cleanup and
you can now type 'make test' in the top level
directory.  'make runtest' will drop you into an
interactive shell.

It could do with some improvements: 'exit' should
probably trigger an orderly shutdown, and signals
aren't handled.

Unfortunately ns_geturl is spectacularly useless,
simply fetching the bytes of a file directly from the
file system.  Is this useful?  I would have expected it
to pass through registered procs etc.

I'm thinking maybe the way to go is an nstest module
which adds some extra commands to dummy up a connection
and run the registered proc.  It would be possible I
guess to also create nstestdb to create a dummy db
driver, and so on...


Feedback appreciated.



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1171843&group_id=130646

[ naviserver-Feature Requests-1159471 ] Virtual Hosting

[SourceForge.net <no...@so...>]

Feature Requests item #1159471, was opened at 2005-03-09 00:40
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159471&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: Virtual Hosting

Initial Comment:
Okay, i did some digging and testing, looks working and
much simpler. I tried to simplify default AS 4.x
virtual hosting and added pageroot virtual hosting,
simple way to use different pagetroots on the same server.

The change vor default virtual hosting is: no
defaultserver anymore, the server who registered
virtual hosts is default server, so nssock is loaded in
the default server, other than that virtual servers are
defined the same way.

One thing i left is to chamge ns_info pageroot to use
Ns_GetConn() and then if exists use connPtr->pageroot,
but this is simple change if you approve current
virtual hosting patch.

Here is the nsd.tcl config example:

ns_section      "ns/server/${server}/module/nssock/servers"
ns_param        test                    vlad.seryakov.com
ns_param        test                   
vlad.seryakov.com:80

ns_section     
"ns/server/${server}/module/nssock/pageroots"
ns_param        ${home}/html/test       vlad.seryakov.com
ns_param        ${home}/html/test      
vlad.seryakov.com:80



----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-28 01:59

Message:
Logged In: YES 
user_id=184124

One notice, hostpreix bt default is vhost, not empty, would
be better to have it empty.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-28 01:55

Message:
Logged In: YES 
user_id=184124

Yes, this worked for me, if Zoran does not mind, you can
commit it.


----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-27 14:25

Message:
Logged In: YES 
user_id=87254


The new serverdir is for storing files associated with a
server (or virtual host) which are not made public via the
pagedir.  So, things like uploads, auto-generated data,
caches etc.  This will contain only the pages directory by
default, because the server doesn't do anything interesting
by default.

I've attached a new patch which makes the serverdir and
pagedir fully customisable. Here are the possible config
settings:

ns_section "ns/server/${servername}/fastpath"
ns_param   serverdir    /srv # default:
$home/servers/$servername
ns_param   pagedir      pages
#ns_param   pageroot    pages # depreciated

ns_section "ns/server/${servername}/vhost"
ns_param   enabled    false
ns_param   hostprefix  vhosts #default ""
ns_param   hosthashlevel 3
ns_param   stripport   true
ns_param   stripwww   true


I've added path hashing, which hashes the hostname so that
they are spread accross multiple directories. e.g.

  foo.com -> ./f/o/o/pages
  a.com -> ./a/_/_/pages

I've also added a -host switch to the ns_serverpath and
ns_pagepath commands so that tool can be written which setup
the necessary directories.


Does this look ok?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 04:26

Message:
Logged In: YES 
user_id=184124

On the other hand if it will be possible to configure/rename
"servers/" and "pages/" parts or omit them completely i
would be happy with this patch.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-14 14:48

Message:
Logged In: YES 
user_id=184124

The problem i see with this patch, it will break all my AS
installations, because it assumes that every servers is under
servers/ direcotory and pageroot is awlays named pages/.

I my on-server installs i do not have this complex directory
structure, i have html/ under ns_info home and pageroot is
set in the nsd.tcl.

Also, what is the point of ServerRoot dir, we have PageRoot
already, if it is set into absolute path other apps use
ns_info pageroot? As i understand, under ServerRoot there is
only pages difrectory? Why not to use pageroot directly?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-14 14:05

Message:
Logged In: YES 
user_id=87254


Oh, forgot to mention.  It depends on the Tcl Callbacks patch.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-14 14:03

Message:
Logged In: YES 
user_id=87254


I've added the patch naviserver-4.0.10-server-root-vhost.patch.

It adds the following new routines:

  Ns_ServerPath()
  Ns_PagePath()
  Ns_SetServerRootProc()
  Ns_ConnLocationAppend()
  Ns_SetConnLocationProc()

  NsServerRoot()
  NsPageRoot()

  ns_serverpath ?pathSegment ...?
  ns_pagepath ?pathSegment ...?
  ns_serverrootproc script ?arg?
  ns_locationproc script ?arg?

And the following new configuration options:

  ns_section "ns/server/${servername}/vhost"
  ns_param   enabled false
  ns_param   prefix ""
  ns_param   pagedir pages # overides fastpath/pageroot
  ns_param   stripwww true
  ns_parma   stripport true

This version of host header based virtual hosting which
depends on the existance of the pages directory is a
superset of the functionality provided by a static
configuration.

Applications which call Ns_ConnLocation() will need to be
modified to call Ns_ConnLocationAppend() or they will not be
vhost aware.

I've reinstated the depreciated proc
Ns_SetConnLocationProc() and changed it's signature.  It
dissapeared from ns.h almost 5 years ago, then reappreared
~2.5 years ago.  I don't think this will be a problem.  In
turn, I've depreciated the Ns_SetLocationProc().

It compiles and the server runs.  I haven't had time to test
it.  Does this look acceptable?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 00:41

Message:
Logged In: YES 
user_id=184124

Just to consider the possibility, instead of mallocing
pageroot/location, by default it can use
sockPtr->pageroot/sockPtr->location, then when ns_conn
pageroot newpageroot called, it will set connPtr->pageroot
with malloced string and ns_conn pageroot will check and
return it instead of sockPtr->pageroot. This way, no
overhead at conn queue and still new pageroot/location can
be set in Tcl

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 00:02

Message:
Logged In: YES 
user_id=184124

That is the probem with nssock, it is actually an extender
of the driver but still somehow kept independent. nssock
itself is useless without driver and only used to bind to
more than one address for different servers. it could be
moved in the core but the problem will be how to define more
than once instance.
malloc are overhead indeed, but once copied they can be used
independently  and canbe set in Tcl by using ns_conn
location newLocation or
ns_conn pageroot newPageroot. In this case they should be a
copies.

Just do not make mass virtual hosting the only virtual
hosting way, being able to change pageroot in the Tcl/C give
developer more flexibility if required. For simple cases,
mass virtual hosting is okay.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-12 23:48

Message:
Logged In: YES 
user_id=87254


I'm not keen at all on adding virtual hosting to the nssock
driver.  There's nothing HTTP specific about the nssock
dirver and I'd like to keep it that way.

There are a couple of problems with the other proposed
solution.  The paired functions
Ns_SetLocationProc()/Ns_ConnSetLocationProc() etc. seems
excessive, and the enforced malloc()ing at runtime of the
location and pageroot strings is an unwelcome overhead.

Using Tls storage is clever but pretty ugly.  I think
dstrings are the way to go here.  I'm not sure Tls is safe
in this implementation.  The same dstring is used for
location and pageroot strings, so it depends what the caller
decides to do with the result and in what order, whether or
not one overwrites the other.

I would like to explore adding mas virtual hosting into the
core.  Let me work up a patch, I think I can get to this
this weekend...



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 21:57

Message:
Logged In: YES 
user_id=184124

Attached is nssock with virtual hosting patch

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 20:39

Message:
Logged In: YES 
user_id=184124

Actually, vhost can be combined with nssock, if options
given it will enable virtual hosting, if not works as
regular sock driver.
This way it is always with the core and at the same time
independent.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 21:38

Message:
Logged In: YES 
user_id=184124

If loaded, vhost module works as Stephan suggested, strips
port/host and usesd pageroot if other root is not specified.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 21:37

Message:
Logged In: YES 
user_id=184124

There is new patch with Stephans corrections/additions.
I think we can provide core module for virtual hosting, i
called it nsvhost and we can extend this module to do all
sorts of hosting.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 18:53

Message:
Logged In: YES 
user_id=184124

And keeping port and www. is sometimes necessary, you can do
virtual hosting by port only, IDT does that for example,
and many sites work without www. prefix, just stripping them
by default may not be appropriate.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 18:24

Message:
Logged In: YES 
user_id=184124

>Looks good.  However, isn't ns_conn pageroot >unneccessary?

For templating engines ns_info pageroot is the only way to
figure out the root, UrlToFile works for fastpath only.

>Would it be better to have the servers and pageroot config
>in one section, to avoid duplication?

They are mutually exclusive, that's why i put them in
different sections, if full virtual server is set, pageroots
are ignored, this is for AOL like virtual hosting with
different servers(rare situation though). So in most cases
pageroots will be used only, thus only one simple syntax.

As for PageRoot and ServerRoot, i think this is a little
cofusing. Currently, pageroot returns full path and whoever
calls pageroot assumes that it will return full path. 

Virtual hosting using directories as hostnames is what i am
currently using with vhost module and i think it can be
included as a standard feature for easy virtualhosting
solutions. 
I do not think this should be the ONLY virtual hosting
solution, everybody can write their own modules using
SetLocation/SetPageRoot procs or register filter which will
set pageroot for each connection.

Let me prepare another patch-set with your suggestions
included .

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-10 08:00

Message:
Logged In: YES 
user_id=87254


Looks good.  However, isn't ns_conn pageroot unneccessary?

Would it be better to have the servers and pageroot config
in one section, to avoid duplication?

ns_section "ns/server/${server}/module/nssock/servers"
ns_param example.com        exampleserver
ns_param foo.example.com   "exampleserver ${home}/whatever"

Here, the first entry maps the example.com host to the
exampleserver and uses it's default pageroot.  The second
entry supplies a new pageroot.


How about mass virtual hosting, i.e. where you don't have to
explicitly configure each host header to pageroot mapping,
but construct the pageroot from the host heafer at runtime?

I've attached the file nsmassvhost.c which implements the
above.  It uses the hooks Ns_PageRootProc and
Ns_LocationProc which would be unneccessary if the
functionality was included as standard.

It trims the port and any leading 'www.' from the host
header.  It would be nice to have this for the static
mapping also, as at the moment to be robust you often need 4
mappings for each virtual host.

It also uses the function Ns_ServerPath().  The idea here is
to introduce the concept of the virtual server root as a
distinct location in the file system, where the pageroot is
a location below that.

I want to change this so that the serverroot is dynamic and
based on the host header (when configured), and the pageroot
is simply the serverroot with "/pages" (or whatever is
configured) appended.

It would look something like this:

/srv/server1/pages
/srv/server1/example.com/pages
/srv/server1/example.com/cache

The first path is the default or non-virtual hosted case. 
server1 is a server defind in the config file and has it's
own private tcl library.  The second path is the pageroot of
a virtual host.  The third is an example of some data which
is specific to the example.com virtual host.

So, without virtual hosts:
  Ns_ServerRoot() -> /srv/server1
  Ns_PageRoot() -> /srv/server1/pages

With virtual hosts (and called in the context of a conn thread):
  Ns_ServerRoot() -> /srv/server1/example.com
  Ns_PageRoot() -> /srv/server1/example.com/pages

The advantage of this system is that you don't have to
restart your server every time you add or remove a virtual
host.  There is also a convenient location to store data
associated with both virtual servers and virtual hosts. 
Easy to backup, remove, etc.

I haven't had time to look at how this would be integrated
into what you've got here, maybe at the weekend.  Feel free
to take a shot at though :-)

Does the scheme outlined above make sense to you?




----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159471&group_id=130646

[ naviserver-Feature Requests-1159471 ] Virtual Hosting

[SourceForge.net <no...@so...>]

Feature Requests item #1159471, was opened at 2005-03-09 00:40
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159471&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: Virtual Hosting

Initial Comment:
Okay, i did some digging and testing, looks working and
much simpler. I tried to simplify default AS 4.x
virtual hosting and added pageroot virtual hosting,
simple way to use different pagetroots on the same server.

The change vor default virtual hosting is: no
defaultserver anymore, the server who registered
virtual hosts is default server, so nssock is loaded in
the default server, other than that virtual servers are
defined the same way.

One thing i left is to chamge ns_info pageroot to use
Ns_GetConn() and then if exists use connPtr->pageroot,
but this is simple change if you approve current
virtual hosting patch.

Here is the nsd.tcl config example:

ns_section      "ns/server/${server}/module/nssock/servers"
ns_param        test                    vlad.seryakov.com
ns_param        test                   
vlad.seryakov.com:80

ns_section     
"ns/server/${server}/module/nssock/pageroots"
ns_param        ${home}/html/test       vlad.seryakov.com
ns_param        ${home}/html/test      
vlad.seryakov.com:80



----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-28 01:55

Message:
Logged In: YES 
user_id=184124

Yes, this worked for me, if Zoran does not mind, you can
commit it.


----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-27 14:25

Message:
Logged In: YES 
user_id=87254


The new serverdir is for storing files associated with a
server (or virtual host) which are not made public via the
pagedir.  So, things like uploads, auto-generated data,
caches etc.  This will contain only the pages directory by
default, because the server doesn't do anything interesting
by default.

I've attached a new patch which makes the serverdir and
pagedir fully customisable. Here are the possible config
settings:

ns_section "ns/server/${servername}/fastpath"
ns_param   serverdir    /srv # default:
$home/servers/$servername
ns_param   pagedir      pages
#ns_param   pageroot    pages # depreciated

ns_section "ns/server/${servername}/vhost"
ns_param   enabled    false
ns_param   hostprefix  vhosts #default ""
ns_param   hosthashlevel 3
ns_param   stripport   true
ns_param   stripwww   true


I've added path hashing, which hashes the hostname so that
they are spread accross multiple directories. e.g.

  foo.com -> ./f/o/o/pages
  a.com -> ./a/_/_/pages

I've also added a -host switch to the ns_serverpath and
ns_pagepath commands so that tool can be written which setup
the necessary directories.


Does this look ok?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 04:26

Message:
Logged In: YES 
user_id=184124

On the other hand if it will be possible to configure/rename
"servers/" and "pages/" parts or omit them completely i
would be happy with this patch.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-14 14:48

Message:
Logged In: YES 
user_id=184124

The problem i see with this patch, it will break all my AS
installations, because it assumes that every servers is under
servers/ direcotory and pageroot is awlays named pages/.

I my on-server installs i do not have this complex directory
structure, i have html/ under ns_info home and pageroot is
set in the nsd.tcl.

Also, what is the point of ServerRoot dir, we have PageRoot
already, if it is set into absolute path other apps use
ns_info pageroot? As i understand, under ServerRoot there is
only pages difrectory? Why not to use pageroot directly?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-14 14:05

Message:
Logged In: YES 
user_id=87254


Oh, forgot to mention.  It depends on the Tcl Callbacks patch.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-14 14:03

Message:
Logged In: YES 
user_id=87254


I've added the patch naviserver-4.0.10-server-root-vhost.patch.

It adds the following new routines:

  Ns_ServerPath()
  Ns_PagePath()
  Ns_SetServerRootProc()
  Ns_ConnLocationAppend()
  Ns_SetConnLocationProc()

  NsServerRoot()
  NsPageRoot()

  ns_serverpath ?pathSegment ...?
  ns_pagepath ?pathSegment ...?
  ns_serverrootproc script ?arg?
  ns_locationproc script ?arg?

And the following new configuration options:

  ns_section "ns/server/${servername}/vhost"
  ns_param   enabled false
  ns_param   prefix ""
  ns_param   pagedir pages # overides fastpath/pageroot
  ns_param   stripwww true
  ns_parma   stripport true

This version of host header based virtual hosting which
depends on the existance of the pages directory is a
superset of the functionality provided by a static
configuration.

Applications which call Ns_ConnLocation() will need to be
modified to call Ns_ConnLocationAppend() or they will not be
vhost aware.

I've reinstated the depreciated proc
Ns_SetConnLocationProc() and changed it's signature.  It
dissapeared from ns.h almost 5 years ago, then reappreared
~2.5 years ago.  I don't think this will be a problem.  In
turn, I've depreciated the Ns_SetLocationProc().

It compiles and the server runs.  I haven't had time to test
it.  Does this look acceptable?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 00:41

Message:
Logged In: YES 
user_id=184124

Just to consider the possibility, instead of mallocing
pageroot/location, by default it can use
sockPtr->pageroot/sockPtr->location, then when ns_conn
pageroot newpageroot called, it will set connPtr->pageroot
with malloced string and ns_conn pageroot will check and
return it instead of sockPtr->pageroot. This way, no
overhead at conn queue and still new pageroot/location can
be set in Tcl

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 00:02

Message:
Logged In: YES 
user_id=184124

That is the probem with nssock, it is actually an extender
of the driver but still somehow kept independent. nssock
itself is useless without driver and only used to bind to
more than one address for different servers. it could be
moved in the core but the problem will be how to define more
than once instance.
malloc are overhead indeed, but once copied they can be used
independently  and canbe set in Tcl by using ns_conn
location newLocation or
ns_conn pageroot newPageroot. In this case they should be a
copies.

Just do not make mass virtual hosting the only virtual
hosting way, being able to change pageroot in the Tcl/C give
developer more flexibility if required. For simple cases,
mass virtual hosting is okay.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-12 23:48

Message:
Logged In: YES 
user_id=87254


I'm not keen at all on adding virtual hosting to the nssock
driver.  There's nothing HTTP specific about the nssock
dirver and I'd like to keep it that way.

There are a couple of problems with the other proposed
solution.  The paired functions
Ns_SetLocationProc()/Ns_ConnSetLocationProc() etc. seems
excessive, and the enforced malloc()ing at runtime of the
location and pageroot strings is an unwelcome overhead.

Using Tls storage is clever but pretty ugly.  I think
dstrings are the way to go here.  I'm not sure Tls is safe
in this implementation.  The same dstring is used for
location and pageroot strings, so it depends what the caller
decides to do with the result and in what order, whether or
not one overwrites the other.

I would like to explore adding mas virtual hosting into the
core.  Let me work up a patch, I think I can get to this
this weekend...



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 21:57

Message:
Logged In: YES 
user_id=184124

Attached is nssock with virtual hosting patch

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 20:39

Message:
Logged In: YES 
user_id=184124

Actually, vhost can be combined with nssock, if options
given it will enable virtual hosting, if not works as
regular sock driver.
This way it is always with the core and at the same time
independent.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 21:38

Message:
Logged In: YES 
user_id=184124

If loaded, vhost module works as Stephan suggested, strips
port/host and usesd pageroot if other root is not specified.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 21:37

Message:
Logged In: YES 
user_id=184124

There is new patch with Stephans corrections/additions.
I think we can provide core module for virtual hosting, i
called it nsvhost and we can extend this module to do all
sorts of hosting.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 18:53

Message:
Logged In: YES 
user_id=184124

And keeping port and www. is sometimes necessary, you can do
virtual hosting by port only, IDT does that for example,
and many sites work without www. prefix, just stripping them
by default may not be appropriate.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 18:24

Message:
Logged In: YES 
user_id=184124

>Looks good.  However, isn't ns_conn pageroot >unneccessary?

For templating engines ns_info pageroot is the only way to
figure out the root, UrlToFile works for fastpath only.

>Would it be better to have the servers and pageroot config
>in one section, to avoid duplication?

They are mutually exclusive, that's why i put them in
different sections, if full virtual server is set, pageroots
are ignored, this is for AOL like virtual hosting with
different servers(rare situation though). So in most cases
pageroots will be used only, thus only one simple syntax.

As for PageRoot and ServerRoot, i think this is a little
cofusing. Currently, pageroot returns full path and whoever
calls pageroot assumes that it will return full path. 

Virtual hosting using directories as hostnames is what i am
currently using with vhost module and i think it can be
included as a standard feature for easy virtualhosting
solutions. 
I do not think this should be the ONLY virtual hosting
solution, everybody can write their own modules using
SetLocation/SetPageRoot procs or register filter which will
set pageroot for each connection.

Let me prepare another patch-set with your suggestions
included .

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-10 08:00

Message:
Logged In: YES 
user_id=87254


Looks good.  However, isn't ns_conn pageroot unneccessary?

Would it be better to have the servers and pageroot config
in one section, to avoid duplication?

ns_section "ns/server/${server}/module/nssock/servers"
ns_param example.com        exampleserver
ns_param foo.example.com   "exampleserver ${home}/whatever"

Here, the first entry maps the example.com host to the
exampleserver and uses it's default pageroot.  The second
entry supplies a new pageroot.


How about mass virtual hosting, i.e. where you don't have to
explicitly configure each host header to pageroot mapping,
but construct the pageroot from the host heafer at runtime?

I've attached the file nsmassvhost.c which implements the
above.  It uses the hooks Ns_PageRootProc and
Ns_LocationProc which would be unneccessary if the
functionality was included as standard.

It trims the port and any leading 'www.' from the host
header.  It would be nice to have this for the static
mapping also, as at the moment to be robust you often need 4
mappings for each virtual host.

It also uses the function Ns_ServerPath().  The idea here is
to introduce the concept of the virtual server root as a
distinct location in the file system, where the pageroot is
a location below that.

I want to change this so that the serverroot is dynamic and
based on the host header (when configured), and the pageroot
is simply the serverroot with "/pages" (or whatever is
configured) appended.

It would look something like this:

/srv/server1/pages
/srv/server1/example.com/pages
/srv/server1/example.com/cache

The first path is the default or non-virtual hosted case. 
server1 is a server defind in the config file and has it's
own private tcl library.  The second path is the pageroot of
a virtual host.  The third is an example of some data which
is specific to the example.com virtual host.

So, without virtual hosts:
  Ns_ServerRoot() -> /srv/server1
  Ns_PageRoot() -> /srv/server1/pages

With virtual hosts (and called in the context of a conn thread):
  Ns_ServerRoot() -> /srv/server1/example.com
  Ns_PageRoot() -> /srv/server1/example.com/pages

The advantage of this system is that you don't have to
restart your server every time you add or remove a virtual
host.  There is also a convenient location to store data
associated with both virtual servers and virtual hosts. 
Easy to backup, remove, etc.

I haven't had time to look at how this would be integrated
into what you've got here, maybe at the weekend.  Feel free
to take a shot at though :-)

Does the scheme outlined above make sense to you?




----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159471&group_id=130646

[ naviserver-Feature Requests-1161597 ] Extend ns_info with info about traces/filters/procs

[SourceForge.net <no...@so...>]

Feature Requests item #1161597, was opened at 2005-03-11 12:30
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1161597&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: Extend ns_info with info about traces/filters/procs

Initial Comment:
Attached is patch that extends ns_info with 3 commands:

ns_info traces
ns_info filters
ns_info requestprocs

They work the same way as ns_info callbacks, no
functionality changed or added, just information commands

----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-27 07:40

Message:
Logged In: YES 
user_id=87254


Hmm, I think the arg proc stuff is just for the callback and
arg registered at some point in the data structure.  The
function which navigates the data structure obviously has to
know the details, so it doesn't need a callback.

So, the only functions which need this treatment would be
NsFastGet and NsAdpProc.  The arg passed to these is really
dull, just a pointer to the server which you'll already know
from walking the url trie, so really all that needs added to
proc.c is something like:

{(void *) NsAdpProc, "ns:adp", NULL},
{(void *) NsFastGet, "ns:fastget", NULL},

i.e. name the proc, but don't bother providing a callback to
describe the arg.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 07:22

Message:
Logged In: YES 
user_id=184124

Yes, this is what i did with UrlSpecific walk function, it
will call Ns_ArpProc for every node which will append
what-ever info to DString. We can live with that for the
beginning, anyway, that info will be used in Tcl and should
be converted into strings. 
So, i guess i can add info proc for UrlSpecific trie?

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-17 04:06

Message:
Logged In: YES 
user_id=87254


To get info on the binary callbacks we're just going to have
to register proc info for all the default filters, request
procs etc.  If we cover everything there should be a
reasonable number of examples for module writers to follow.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 18:47

Message:
Logged In: YES 
user_id=184124

Also, somein docs would be usefull to put info how to get
info about some binary callbacks:

Welcome to ossweb running at /usr/local/ns/bin/nsd (pid 5524)
ossweb:nscp 1> ns_info traces
p:0xb7187be0 a:0x80817d8

# gdb
(gdb) attach 12345

(gdb) info symbol 0xb7187be0
LogTrace in section .text





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 18:44

Message:
Logged In: YES 
user_id=184124

I started with minimal changes, #defines and &syntax can be
chnages of course. now that you started TclCallbacks, this
thing depends when you commit your changes, it requires
Ns_ArgProc and ns_info stuff.
Once we have unified API for callbask i will change it to be
server-specific as well, it took a while to figure out how
it works :-))

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-12 18:37

Message:
Logged In: YES 
user_id=87254


This is a nice addition.  

ns_info filters/traces is server specific, but ns_info
requestprocs returns info for all servers.  Can you make
this per-server also, to be consistent?

The URL walking procs look a little tricky, I tnink they
would benefit from having their own more detailed comments.

The comment currently says it calls a function for each
node, and the effect depends on the function.  But the
function signature is Ns_ArgProc so there's really only one
thing it can do.

Why does it fix the stack size at 512?  Can these magic
numbers be #define'd.

Using a while loop rather than an empty for loop would be a
little easier to read.

There's a lot of this kind of thing:
    (&(triePtr->branches))->n
Could this be simplified to:
    triePtr->branches.n ?


Otherwise, looks pretty good.  I like it.



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1161597&group_id=130646

[ naviserver-Feature Requests-1159471 ] Virtual Hosting

[SourceForge.net <no...@so...>]

Feature Requests item #1159471, was opened at 2005-03-08 17:40
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159471&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: Virtual Hosting

Initial Comment:
Okay, i did some digging and testing, looks working and
much simpler. I tried to simplify default AS 4.x
virtual hosting and added pageroot virtual hosting,
simple way to use different pagetroots on the same server.

The change vor default virtual hosting is: no
defaultserver anymore, the server who registered
virtual hosts is default server, so nssock is loaded in
the default server, other than that virtual servers are
defined the same way.

One thing i left is to chamge ns_info pageroot to use
Ns_GetConn() and then if exists use connPtr->pageroot,
but this is simple change if you approve current
virtual hosting patch.

Here is the nsd.tcl config example:

ns_section      "ns/server/${server}/module/nssock/servers"
ns_param        test                    vlad.seryakov.com
ns_param        test                   
vlad.seryakov.com:80

ns_section     
"ns/server/${server}/module/nssock/pageroots"
ns_param        ${home}/html/test       vlad.seryakov.com
ns_param        ${home}/html/test      
vlad.seryakov.com:80



----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-27 07:25

Message:
Logged In: YES 
user_id=87254


The new serverdir is for storing files associated with a
server (or virtual host) which are not made public via the
pagedir.  So, things like uploads, auto-generated data,
caches etc.  This will contain only the pages directory by
default, because the server doesn't do anything interesting
by default.

I've attached a new patch which makes the serverdir and
pagedir fully customisable. Here are the possible config
settings:

ns_section "ns/server/${servername}/fastpath"
ns_param   serverdir    /srv # default:
$home/servers/$servername
ns_param   pagedir      pages
#ns_param   pageroot    pages # depreciated

ns_section "ns/server/${servername}/vhost"
ns_param   enabled    false
ns_param   hostprefix  vhosts #default ""
ns_param   hosthashlevel 3
ns_param   stripport   true
ns_param   stripwww   true


I've added path hashing, which hashes the hostname so that
they are spread accross multiple directories. e.g.

  foo.com -> ./f/o/o/pages
  a.com -> ./a/_/_/pages

I've also added a -host switch to the ns_serverpath and
ns_pagepath commands so that tool can be written which setup
the necessary directories.


Does this look ok?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 21:26

Message:
Logged In: YES 
user_id=184124

On the other hand if it will be possible to configure/rename
"servers/" and "pages/" parts or omit them completely i
would be happy with this patch.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-14 07:48

Message:
Logged In: YES 
user_id=184124

The problem i see with this patch, it will break all my AS
installations, because it assumes that every servers is under
servers/ direcotory and pageroot is awlays named pages/.

I my on-server installs i do not have this complex directory
structure, i have html/ under ns_info home and pageroot is
set in the nsd.tcl.

Also, what is the point of ServerRoot dir, we have PageRoot
already, if it is set into absolute path other apps use
ns_info pageroot? As i understand, under ServerRoot there is
only pages difrectory? Why not to use pageroot directly?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-14 07:05

Message:
Logged In: YES 
user_id=87254


Oh, forgot to mention.  It depends on the Tcl Callbacks patch.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-14 07:03

Message:
Logged In: YES 
user_id=87254


I've added the patch naviserver-4.0.10-server-root-vhost.patch.

It adds the following new routines:

  Ns_ServerPath()
  Ns_PagePath()
  Ns_SetServerRootProc()
  Ns_ConnLocationAppend()
  Ns_SetConnLocationProc()

  NsServerRoot()
  NsPageRoot()

  ns_serverpath ?pathSegment ...?
  ns_pagepath ?pathSegment ...?
  ns_serverrootproc script ?arg?
  ns_locationproc script ?arg?

And the following new configuration options:

  ns_section "ns/server/${servername}/vhost"
  ns_param   enabled false
  ns_param   prefix ""
  ns_param   pagedir pages # overides fastpath/pageroot
  ns_param   stripwww true
  ns_parma   stripport true

This version of host header based virtual hosting which
depends on the existance of the pages directory is a
superset of the functionality provided by a static
configuration.

Applications which call Ns_ConnLocation() will need to be
modified to call Ns_ConnLocationAppend() or they will not be
vhost aware.

I've reinstated the depreciated proc
Ns_SetConnLocationProc() and changed it's signature.  It
dissapeared from ns.h almost 5 years ago, then reappreared
~2.5 years ago.  I don't think this will be a problem.  In
turn, I've depreciated the Ns_SetLocationProc().

It compiles and the server runs.  I haven't had time to test
it.  Does this look acceptable?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 17:41

Message:
Logged In: YES 
user_id=184124

Just to consider the possibility, instead of mallocing
pageroot/location, by default it can use
sockPtr->pageroot/sockPtr->location, then when ns_conn
pageroot newpageroot called, it will set connPtr->pageroot
with malloced string and ns_conn pageroot will check and
return it instead of sockPtr->pageroot. This way, no
overhead at conn queue and still new pageroot/location can
be set in Tcl

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 17:02

Message:
Logged In: YES 
user_id=184124

That is the probem with nssock, it is actually an extender
of the driver but still somehow kept independent. nssock
itself is useless without driver and only used to bind to
more than one address for different servers. it could be
moved in the core but the problem will be how to define more
than once instance.
malloc are overhead indeed, but once copied they can be used
independently  and canbe set in Tcl by using ns_conn
location newLocation or
ns_conn pageroot newPageroot. In this case they should be a
copies.

Just do not make mass virtual hosting the only virtual
hosting way, being able to change pageroot in the Tcl/C give
developer more flexibility if required. For simple cases,
mass virtual hosting is okay.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-12 16:48

Message:
Logged In: YES 
user_id=87254


I'm not keen at all on adding virtual hosting to the nssock
driver.  There's nothing HTTP specific about the nssock
dirver and I'd like to keep it that way.

There are a couple of problems with the other proposed
solution.  The paired functions
Ns_SetLocationProc()/Ns_ConnSetLocationProc() etc. seems
excessive, and the enforced malloc()ing at runtime of the
location and pageroot strings is an unwelcome overhead.

Using Tls storage is clever but pretty ugly.  I think
dstrings are the way to go here.  I'm not sure Tls is safe
in this implementation.  The same dstring is used for
location and pageroot strings, so it depends what the caller
decides to do with the result and in what order, whether or
not one overwrites the other.

I would like to explore adding mas virtual hosting into the
core.  Let me work up a patch, I think I can get to this
this weekend...



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 14:57

Message:
Logged In: YES 
user_id=184124

Attached is nssock with virtual hosting patch

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 13:39

Message:
Logged In: YES 
user_id=184124

Actually, vhost can be combined with nssock, if options
given it will enable virtual hosting, if not works as
regular sock driver.
This way it is always with the core and at the same time
independent.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 14:38

Message:
Logged In: YES 
user_id=184124

If loaded, vhost module works as Stephan suggested, strips
port/host and usesd pageroot if other root is not specified.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 14:37

Message:
Logged In: YES 
user_id=184124

There is new patch with Stephans corrections/additions.
I think we can provide core module for virtual hosting, i
called it nsvhost and we can extend this module to do all
sorts of hosting.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 11:53

Message:
Logged In: YES 
user_id=184124

And keeping port and www. is sometimes necessary, you can do
virtual hosting by port only, IDT does that for example,
and many sites work without www. prefix, just stripping them
by default may not be appropriate.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-10 11:24

Message:
Logged In: YES 
user_id=184124

>Looks good.  However, isn't ns_conn pageroot >unneccessary?

For templating engines ns_info pageroot is the only way to
figure out the root, UrlToFile works for fastpath only.

>Would it be better to have the servers and pageroot config
>in one section, to avoid duplication?

They are mutually exclusive, that's why i put them in
different sections, if full virtual server is set, pageroots
are ignored, this is for AOL like virtual hosting with
different servers(rare situation though). So in most cases
pageroots will be used only, thus only one simple syntax.

As for PageRoot and ServerRoot, i think this is a little
cofusing. Currently, pageroot returns full path and whoever
calls pageroot assumes that it will return full path. 

Virtual hosting using directories as hostnames is what i am
currently using with vhost module and i think it can be
included as a standard feature for easy virtualhosting
solutions. 
I do not think this should be the ONLY virtual hosting
solution, everybody can write their own modules using
SetLocation/SetPageRoot procs or register filter which will
set pageroot for each connection.

Let me prepare another patch-set with your suggestions
included .

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-10 01:00

Message:
Logged In: YES 
user_id=87254


Looks good.  However, isn't ns_conn pageroot unneccessary?

Would it be better to have the servers and pageroot config
in one section, to avoid duplication?

ns_section "ns/server/${server}/module/nssock/servers"
ns_param example.com        exampleserver
ns_param foo.example.com   "exampleserver ${home}/whatever"

Here, the first entry maps the example.com host to the
exampleserver and uses it's default pageroot.  The second
entry supplies a new pageroot.


How about mass virtual hosting, i.e. where you don't have to
explicitly configure each host header to pageroot mapping,
but construct the pageroot from the host heafer at runtime?

I've attached the file nsmassvhost.c which implements the
above.  It uses the hooks Ns_PageRootProc and
Ns_LocationProc which would be unneccessary if the
functionality was included as standard.

It trims the port and any leading 'www.' from the host
header.  It would be nice to have this for the static
mapping also, as at the moment to be robust you often need 4
mappings for each virtual host.

It also uses the function Ns_ServerPath().  The idea here is
to introduce the concept of the virtual server root as a
distinct location in the file system, where the pageroot is
a location below that.

I want to change this so that the serverroot is dynamic and
based on the host header (when configured), and the pageroot
is simply the serverroot with "/pages" (or whatever is
configured) appended.

It would look something like this:

/srv/server1/pages
/srv/server1/example.com/pages
/srv/server1/example.com/cache

The first path is the default or non-virtual hosted case. 
server1 is a server defind in the config file and has it's
own private tcl library.  The second path is the pageroot of
a virtual host.  The third is an example of some data which
is specific to the example.com virtual host.

So, without virtual hosts:
  Ns_ServerRoot() -> /srv/server1
  Ns_PageRoot() -> /srv/server1/pages

With virtual hosts (and called in the context of a conn thread):
  Ns_ServerRoot() -> /srv/server1/example.com
  Ns_PageRoot() -> /srv/server1/example.com/pages

The advantage of this system is that you don't have to
restart your server every time you add or remove a virtual
host.  There is also a convenient location to store data
associated with both virtual servers and virtual hosts. 
Easy to backup, remove, etc.

I haven't had time to look at how this would be integrated
into what you've got here, maybe at the weekend.  Feel free
to take a shot at though :-)

Does the scheme outlined above make sense to you?




----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159471&group_id=130646

[ naviserver-Feature Requests-1162223 ] Tcl Callbacks

[SourceForge.net <no...@so...>]

Feature Requests item #1162223, was opened at 2005-03-13 00:22
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

Category: None
Group: None
Status: Closed
Resolution: Accepted
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: Tcl Callbacks

Initial Comment:

The server offers a dual C/Tcl interface.  There are a
number of of places where Tcl code needs to be executed
by the C core, or the core doesn't know whether a
callback is C or Tcl.

I've attached the patch
naviserver-4.1-tclcallbacks.patch.  It extracts the Tcl
callback infrastructure from tclsched.c into
tclcallbacks.c and makes it a public interface.  Here
it is:

Ns_TclNewCallback()
Ns_TclNewCallbackObj()
Ns_TclEvalCallback()
Ns_TclCallbackProc()
Ns_TclFreeCallback()
Ns_TclCallbackArgProc()

It provides a standard interface for Tcl callbacks
throughout the server, handling interp allocation,
introspection, evaluation and cleanup.

Also, tclcallbacks.c is somewhat analogous to
callbacks.c, and so the generic ns_atexit etc.
callbacks which were in tclsched.c have been moved to
the tclcallbacks file.

The first set of code to be converted to the new
interface is tclsched.c itself, and I've converted to
using Ns_ParseObjv() while I was in there.  File size
has dropped ~250 lines.


One thing I'm wondering about is whether the
Ns_TclEvalCallback() routine should take extra args to
send to the Tcl proc.  There seems to be no standard
scheme for ordering the args so there still needs to be
a way to manually construct the command, which is why
the members of the Ns_TclCallback struct are public. 
But for this very reason, it might be worth providing a
standard way to do this so all future callbacks are
consistent.


It would be nice in the future to come up with some way
of registereing these callback points such that you
don't have to create an ns_info subcommand for example
manually.  It should be possible for loadable modules
to register their callback points and have them show up
in the right place.



----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-26 21:25

Message:
Logged In: YES 
user_id=184124

It is a good start, so now i can port urlspace/trie walk
utility to new Tcl callbacks interface

----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-26 18:28

Message:
Logged In: YES 
user_id=95086

I do not mind at all. It is good to have someting to start with and 
see how it develops.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-26 18:21

Message:
Logged In: YES 
user_id=87254


Yes, accepting an arbitrary number of args is a good idea. 
It does free you from having to use [list ...] to get
variable subsitution etc.

I wasn't able to come up with a grand plan for callbacks so
I've commited the patch with one small change, as there's
other changes that depend on it.  I'd like to get some
experience using it for a while and maybe the grand plan
will become more obvious.  I hope you don't mind.

The change I made was to have Ns_TclEvalCallback() accept
varargs.  It allows the caller to present an interface like:

    atclproc varg1 varg2 ?context?

to the Tcl proc.  The vargs come from the C code, context is
as passed by Tcl when the callback was registered.



----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 18:54

Message:
Logged In: YES 
user_id=95086

After thinking awhile, hm.... why not pass *variable* 
number of arguments to the callback instead of one? 
So: 
 
  register_callback script arg ?arg ...? 
 
and stuff script all args in the callback structure? 
This is ultimate flexibility, or? 
 
Or even go to the other extent and use no arguments at all. 
After all, script is script; you can evaluate just about any 
proper Tcl string. The caller routine may or may not supply 
additional arguments as it usually would (example [fcopy]). 
Hm? Or am I missing the point? 
 
 
 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 14:16

Message:
Logged In: YES 
user_id=184124

Yes, let's start slowly, we know the direction we are going,
let 's do it step by step starting with simple things. We
can have generic Tcl callbacks already and all new callbacks
will use it and we will see how flexible this mechanism is.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-17 11:28

Message:
Logged In: YES 
user_id=87254


Yes, I forgot to mention that Tcl callbacks are a little
more general than those for C in callbacks.c.  Those take a
single arg which is the context you registered with the
proc, and return void.   Tcl callbacks on the other hand can
return a result which at the C level gets appended to a
dstring if you pass that in.

The members of the Ns_TclCallback structure are public so
that you can construct a command and args in any format. 
This is the question I was asking originally: should there
be a default recomended way to do this?

So for example Ns_TclEvalCallback() might take varargs
cstrings, which if passed would be added to the tcl command
to execute:

  command ?cbarg1? ?cbarg2? ?context?

Where cbarg2 and cbarg2 are specific to the type of callback
and context is what was registered with the proc by the Tcl
programmer.

Unfortunately that wouldn't cover all the bases as some of
the existing APIs do weird things like place callback
specific args after the context or if no context was passed,
always append an empty string.  For new callbacks, something
standard like the above scheme should be used.


If you look at the patch for virtual hosting you'll see that
two new Tcl callbacks were created and they contain simillar
boiler plate.  This makes me think that maybe there should
be something like an [ns_addcallback serverroot myrootproc
mycontext] interface.

It's a little more complicated than that because some
callbacks are singular, like the serverroot proc, and some
are plural, like filters.  So maybe the thing to do is get
some experience with this simple version and then expand it
later...





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 00:46

Message:
Logged In: YES 
user_id=184124

It looks good, unified callback interface would be better
but Tcl filters. traces, requestprocs uses different
arguments/parameters and currently the only way to get info
about them to use ns_info interface and Ns_ArgProc
interface. Tcl_Callback are for simple proc/args callbacks,
filters is special case, see if you can convert them.
But i am for this change.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-13 00:30

Message:
Logged In: YES 
user_id=87254


I've attached a patch which converts the tclrequest.c file
to use the new callback API as an example.  It also uses the
Ns_ParseObjv() API, and I've removed support for the legacy
conn variable.

It removes ~200 lines and all utility functions.  The
callback API provides default Ns_ArgProc introspection, but
this will be much nicer when combined with Vlad's RFE
1161597  :-)



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

[ naviserver-Feature Requests-1166553 ] Add ttrace module

[SourceForge.net <no...@so...>]

Feature Requests item #1166553, was opened at 2005-03-19 17:35
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Zoran Vasiljevic (vasiljevic)
Summary: Add ttrace module

Initial Comment:
The ttrace module is written in Tcl and is part of the
threading
extension and available on SF:

  http://sourceforge.net/projects/ttrace

The basic idea is to use Tcl trace capabilities and
replicate state
of the Tcl interpreter from one to other threads in a
lazy mode.
So, instead of loading everything in one interp and
then using clumsy
introspective script to build the initialization script
for other threads
the ttrace package loads all procedure, namespace (et
al) defs
and stuffs them into a shared array. By overloading the
Tcl [unknown]
command, resources are brought into life as needed. 
This results in drastic memory footprint reduction and
much faster
thread creation. 

I have it running for about 1 year w/o any noticable
speed penalty.
One of the OpenACS users also tried it and is very
pleased. I think
we can add this as alternate interp initialization scheme.




----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-26 19:34

Message:
Logged In: YES 
user_id=95086

Yep. It is little bit difficult to follow. I think I will have to add some
more clarifying comments.

But the basic idea is trivial.

Setup Tcl-level traces on various commands (load, proc, namespace..)
and capture the work in nsv_arrays. Then synthethize a small Tcl script
which, among other things, overloads the Tcl [unknown] command.
The script is passed to each interpreter. 
During runtime, the [unknown] is responsible for pulling things out ot the
nsv_arrays and loading them in the running inter.
That's it.
OK, the implementation is not that simple, but this is life. It will be more
clear when I add some comments there...
Bottom line: it works perfectly for me and for  some other guys who dared
to test it :). Overall, reduction in memory was significant (2-3 times) and 
the thread init time drop to couple of msecs. 


----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-26 19:26

Message:
Logged In: YES 
user_id=87254


Good idea.  I remember you mentioning rewriting in C some
time ago, is that still the plan?

I took a quick look at the ttrace module in sf cvs and had a
little trouble following it.  I guess that's just the nature
of the problem, there seems to be a lot of fiddly things to
consider.  It would be nice to make it more obvious what
steps are being taken, but nothing comes to mind immediately...



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-19 21:30

Message:
Logged In: YES 
user_id=184124

could be a good option, yes

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

[ naviserver-Feature Requests-1162223 ] Tcl Callbacks

[SourceForge.net <no...@so...>]

Feature Requests item #1162223, was opened at 2005-03-13 01:22
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

Category: None
Group: None
Status: Closed
Resolution: Accepted
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: Tcl Callbacks

Initial Comment:

The server offers a dual C/Tcl interface.  There are a
number of of places where Tcl code needs to be executed
by the C core, or the core doesn't know whether a
callback is C or Tcl.

I've attached the patch
naviserver-4.1-tclcallbacks.patch.  It extracts the Tcl
callback infrastructure from tclsched.c into
tclcallbacks.c and makes it a public interface.  Here
it is:

Ns_TclNewCallback()
Ns_TclNewCallbackObj()
Ns_TclEvalCallback()
Ns_TclCallbackProc()
Ns_TclFreeCallback()
Ns_TclCallbackArgProc()

It provides a standard interface for Tcl callbacks
throughout the server, handling interp allocation,
introspection, evaluation and cleanup.

Also, tclcallbacks.c is somewhat analogous to
callbacks.c, and so the generic ns_atexit etc.
callbacks which were in tclsched.c have been moved to
the tclcallbacks file.

The first set of code to be converted to the new
interface is tclsched.c itself, and I've converted to
using Ns_ParseObjv() while I was in there.  File size
has dropped ~250 lines.


One thing I'm wondering about is whether the
Ns_TclEvalCallback() routine should take extra args to
send to the Tcl proc.  There seems to be no standard
scheme for ordering the args so there still needs to be
a way to manually construct the command, which is why
the members of the Ns_TclCallback struct are public. 
But for this very reason, it might be worth providing a
standard way to do this so all future callbacks are
consistent.


It would be nice in the future to come up with some way
of registereing these callback points such that you
don't have to create an ns_info subcommand for example
manually.  It should be possible for loadable modules
to register their callback points and have them show up
in the right place.



----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-26 19:28

Message:
Logged In: YES 
user_id=95086

I do not mind at all. It is good to have someting to start with and 
see how it develops.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-26 19:21

Message:
Logged In: YES 
user_id=87254


Yes, accepting an arbitrary number of args is a good idea. 
It does free you from having to use [list ...] to get
variable subsitution etc.

I wasn't able to come up with a grand plan for callbacks so
I've commited the patch with one small change, as there's
other changes that depend on it.  I'd like to get some
experience using it for a while and maybe the grand plan
will become more obvious.  I hope you don't mind.

The change I made was to have Ns_TclEvalCallback() accept
varargs.  It allows the caller to present an interface like:

    atclproc varg1 varg2 ?context?

to the Tcl proc.  The vargs come from the C code, context is
as passed by Tcl when the callback was registered.



----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 19:54

Message:
Logged In: YES 
user_id=95086

After thinking awhile, hm.... why not pass *variable* 
number of arguments to the callback instead of one? 
So: 
 
  register_callback script arg ?arg ...? 
 
and stuff script all args in the callback structure? 
This is ultimate flexibility, or? 
 
Or even go to the other extent and use no arguments at all. 
After all, script is script; you can evaluate just about any 
proper Tcl string. The caller routine may or may not supply 
additional arguments as it usually would (example [fcopy]). 
Hm? Or am I missing the point? 
 
 
 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 15:16

Message:
Logged In: YES 
user_id=184124

Yes, let's start slowly, we know the direction we are going,
let 's do it step by step starting with simple things. We
can have generic Tcl callbacks already and all new callbacks
will use it and we will see how flexible this mechanism is.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-17 12:28

Message:
Logged In: YES 
user_id=87254


Yes, I forgot to mention that Tcl callbacks are a little
more general than those for C in callbacks.c.  Those take a
single arg which is the context you registered with the
proc, and return void.   Tcl callbacks on the other hand can
return a result which at the C level gets appended to a
dstring if you pass that in.

The members of the Ns_TclCallback structure are public so
that you can construct a command and args in any format. 
This is the question I was asking originally: should there
be a default recomended way to do this?

So for example Ns_TclEvalCallback() might take varargs
cstrings, which if passed would be added to the tcl command
to execute:

  command ?cbarg1? ?cbarg2? ?context?

Where cbarg2 and cbarg2 are specific to the type of callback
and context is what was registered with the proc by the Tcl
programmer.

Unfortunately that wouldn't cover all the bases as some of
the existing APIs do weird things like place callback
specific args after the context or if no context was passed,
always append an empty string.  For new callbacks, something
standard like the above scheme should be used.


If you look at the patch for virtual hosting you'll see that
two new Tcl callbacks were created and they contain simillar
boiler plate.  This makes me think that maybe there should
be something like an [ns_addcallback serverroot myrootproc
mycontext] interface.

It's a little more complicated than that because some
callbacks are singular, like the serverroot proc, and some
are plural, like filters.  So maybe the thing to do is get
some experience with this simple version and then expand it
later...





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 01:46

Message:
Logged In: YES 
user_id=184124

It looks good, unified callback interface would be better
but Tcl filters. traces, requestprocs uses different
arguments/parameters and currently the only way to get info
about them to use ns_info interface and Ns_ArgProc
interface. Tcl_Callback are for simple proc/args callbacks,
filters is special case, see if you can convert them.
But i am for this change.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-13 01:30

Message:
Logged In: YES 
user_id=87254


I've attached a patch which converts the tclrequest.c file
to use the new callback API as an example.  It also uses the
Ns_ParseObjv() API, and I've removed support for the legacy
conn variable.

It removes ~200 lines and all utility functions.  The
callback API provides default Ns_ArgProc introspection, but
this will be much nicer when combined with Vlad's RFE
1161597  :-)



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

[ naviserver-Feature Requests-1166553 ] Add ttrace module

[SourceForge.net <no...@so...>]

Feature Requests item #1166553, was opened at 2005-03-19 09:35
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Zoran Vasiljevic (vasiljevic)
Summary: Add ttrace module

Initial Comment:
The ttrace module is written in Tcl and is part of the
threading
extension and available on SF:

  http://sourceforge.net/projects/ttrace

The basic idea is to use Tcl trace capabilities and
replicate state
of the Tcl interpreter from one to other threads in a
lazy mode.
So, instead of loading everything in one interp and
then using clumsy
introspective script to build the initialization script
for other threads
the ttrace package loads all procedure, namespace (et
al) defs
and stuffs them into a shared array. By overloading the
Tcl [unknown]
command, resources are brought into life as needed. 
This results in drastic memory footprint reduction and
much faster
thread creation. 

I have it running for about 1 year w/o any noticable
speed penalty.
One of the OpenACS users also tried it and is very
pleased. I think
we can add this as alternate interp initialization scheme.




----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-26 11:26

Message:
Logged In: YES 
user_id=87254


Good idea.  I remember you mentioning rewriting in C some
time ago, is that still the plan?

I took a quick look at the ttrace module in sf cvs and had a
little trouble following it.  I guess that's just the nature
of the problem, there seems to be a lot of fiddly things to
consider.  It would be nice to make it more obvious what
steps are being taken, but nothing comes to mind immediately...



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-19 13:30

Message:
Logged In: YES 
user_id=184124

could be a good option, yes

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

[ naviserver-Feature Requests-1162223 ] Tcl Callbacks

[SourceForge.net <no...@so...>]

Feature Requests item #1162223, was opened at 2005-03-12 17:22
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

Category: None
Group: None
>Status: Closed
>Resolution: Accepted
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: Tcl Callbacks

Initial Comment:

The server offers a dual C/Tcl interface.  There are a
number of of places where Tcl code needs to be executed
by the C core, or the core doesn't know whether a
callback is C or Tcl.

I've attached the patch
naviserver-4.1-tclcallbacks.patch.  It extracts the Tcl
callback infrastructure from tclsched.c into
tclcallbacks.c and makes it a public interface.  Here
it is:

Ns_TclNewCallback()
Ns_TclNewCallbackObj()
Ns_TclEvalCallback()
Ns_TclCallbackProc()
Ns_TclFreeCallback()
Ns_TclCallbackArgProc()

It provides a standard interface for Tcl callbacks
throughout the server, handling interp allocation,
introspection, evaluation and cleanup.

Also, tclcallbacks.c is somewhat analogous to
callbacks.c, and so the generic ns_atexit etc.
callbacks which were in tclsched.c have been moved to
the tclcallbacks file.

The first set of code to be converted to the new
interface is tclsched.c itself, and I've converted to
using Ns_ParseObjv() while I was in there.  File size
has dropped ~250 lines.


One thing I'm wondering about is whether the
Ns_TclEvalCallback() routine should take extra args to
send to the Tcl proc.  There seems to be no standard
scheme for ordering the args so there still needs to be
a way to manually construct the command, which is why
the members of the Ns_TclCallback struct are public. 
But for this very reason, it might be worth providing a
standard way to do this so all future callbacks are
consistent.


It would be nice in the future to come up with some way
of registereing these callback points such that you
don't have to create an ns_info subcommand for example
manually.  It should be possible for loadable modules
to register their callback points and have them show up
in the right place.



----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-26 11:21

Message:
Logged In: YES 
user_id=87254


Yes, accepting an arbitrary number of args is a good idea. 
It does free you from having to use [list ...] to get
variable subsitution etc.

I wasn't able to come up with a grand plan for callbacks so
I've commited the patch with one small change, as there's
other changes that depend on it.  I'd like to get some
experience using it for a while and maybe the grand plan
will become more obvious.  I hope you don't mind.

The change I made was to have Ns_TclEvalCallback() accept
varargs.  It allows the caller to present an interface like:

    atclproc varg1 varg2 ?context?

to the Tcl proc.  The vargs come from the C code, context is
as passed by Tcl when the callback was registered.



----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 11:54

Message:
Logged In: YES 
user_id=95086

After thinking awhile, hm.... why not pass *variable* 
number of arguments to the callback instead of one? 
So: 
 
  register_callback script arg ?arg ...? 
 
and stuff script all args in the callback structure? 
This is ultimate flexibility, or? 
 
Or even go to the other extent and use no arguments at all. 
After all, script is script; you can evaluate just about any 
proper Tcl string. The caller routine may or may not supply 
additional arguments as it usually would (example [fcopy]). 
Hm? Or am I missing the point? 
 
 
 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 07:16

Message:
Logged In: YES 
user_id=184124

Yes, let's start slowly, we know the direction we are going,
let 's do it step by step starting with simple things. We
can have generic Tcl callbacks already and all new callbacks
will use it and we will see how flexible this mechanism is.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-17 04:28

Message:
Logged In: YES 
user_id=87254


Yes, I forgot to mention that Tcl callbacks are a little
more general than those for C in callbacks.c.  Those take a
single arg which is the context you registered with the
proc, and return void.   Tcl callbacks on the other hand can
return a result which at the C level gets appended to a
dstring if you pass that in.

The members of the Ns_TclCallback structure are public so
that you can construct a command and args in any format. 
This is the question I was asking originally: should there
be a default recomended way to do this?

So for example Ns_TclEvalCallback() might take varargs
cstrings, which if passed would be added to the tcl command
to execute:

  command ?cbarg1? ?cbarg2? ?context?

Where cbarg2 and cbarg2 are specific to the type of callback
and context is what was registered with the proc by the Tcl
programmer.

Unfortunately that wouldn't cover all the bases as some of
the existing APIs do weird things like place callback
specific args after the context or if no context was passed,
always append an empty string.  For new callbacks, something
standard like the above scheme should be used.


If you look at the patch for virtual hosting you'll see that
two new Tcl callbacks were created and they contain simillar
boiler plate.  This makes me think that maybe there should
be something like an [ns_addcallback serverroot myrootproc
mycontext] interface.

It's a little more complicated than that because some
callbacks are singular, like the serverroot proc, and some
are plural, like filters.  So maybe the thing to do is get
some experience with this simple version and then expand it
later...





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-12 17:46

Message:
Logged In: YES 
user_id=184124

It looks good, unified callback interface would be better
but Tcl filters. traces, requestprocs uses different
arguments/parameters and currently the only way to get info
about them to use ns_info interface and Ns_ArgProc
interface. Tcl_Callback are for simple proc/args callbacks,
filters is special case, see if you can convert them.
But i am for this change.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-12 17:30

Message:
Logged In: YES 
user_id=87254


I've attached a patch which converts the tclrequest.c file
to use the new callback API as an example.  It also uses the
Ns_ParseObjv() API, and I've removed support for the legacy
conn variable.

It removes ~200 lines and all utility functions.  The
callback API provides default Ns_ArgProc introspection, but
this will be much nicer when combined with Vlad's RFE
1161597  :-)



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

[ naviserver-Feature Requests-1145957 ] Cookie support for C and Tcl

[SourceForge.net <no...@so...>]

Feature Requests item #1145957, was opened at 2005-02-21 20:13
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1145957&group_id=130646

Category: None
Group: None
>Status: Closed
>Resolution: Accepted
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: Cookie support for C and Tcl

Initial Comment:

Atached is a patch which implements a simple cookie API
for both C and Tcl.  Here's what it looks like (the
formating will probably get mangled, sorry):

void Ns_ConnSetCookie(Ns_Conn *conn,  char *name, char
*value, int maxage);
void Ns_ConnSetSecureCookie(Ns_Conn *conn,  char *name,
char *value, int maxage);
void Ns_ConnSetCookieEx(Ns_Conn *conn,  char *name,
char *value, int maxage, char *domain, char *path, int
secure);
void Ns_ConnDeleteCookie(Ns_Conn *conn, char *name,
char *domain, char *path);
void Ns_ConnDeleteSecureCookie(Ns_Conn *conn, char
*name, char *domain, char *path);
char *Ns_ConnGetCookie(Ns_DString *dest, Ns_Conn *conn,
char *name);

ns_setcookie ?-secure bool? ?-domain d? ?-path p?
?-maxage secs? name data
ns_getcookie name ?default?
ns_deletecookie ?-secure bool? ?-domain d? ?-path p? name


As you can see it only tackles the basics, no comment
or version field.  Although I've never personaly had to
use those features.  

Cookies are a pre-requisite for the authentication
module which exercises the new autht/authz API I
haven't written yet...

Obviously, the string parsing in the GetCookie code
will have to be very carefully looked at before this
sees the Internet.  But in general, does this look like
it's on the right track?



----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-26 08:54

Message:
Logged In: YES 
user_id=87254


Thanks.  I've commited this to cvs.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 21:55

Message:
Logged In: YES 
user_id=184124

I uploaded modified patch which looks into all Cookies headers.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-03 16:25

Message:
Logged In: YES 
user_id=87254


Good call.  I've been using a simillar algorithm in Tcl only
and it works fine, but I guess that's because I've only been
storing session keys etc. which are quite small.  I'll have
to examine all the cookie headers.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 09:05

Message:
Logged In: YES 
user_id=184124

Looks good, the only issue i have, some browser can send
more than one Cookie header and you are looking only in the
first header. Also, this is new API, no problem adding it.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1145957&group_id=130646

[ naviserver-Feature Requests-1166553 ] Add ttrace module

[SourceForge.net <no...@so...>]

Feature Requests item #1166553, was opened at 2005-03-19 16:35
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Zoran Vasiljevic (vasiljevic)
Summary: Add ttrace module

Initial Comment:
The ttrace module is written in Tcl and is part of the
threading
extension and available on SF:

  http://sourceforge.net/projects/ttrace

The basic idea is to use Tcl trace capabilities and
replicate state
of the Tcl interpreter from one to other threads in a
lazy mode.
So, instead of loading everything in one interp and
then using clumsy
introspective script to build the initialization script
for other threads
the ttrace package loads all procedure, namespace (et
al) defs
and stuffs them into a shared array. By overloading the
Tcl [unknown]
command, resources are brought into life as needed. 
This results in drastic memory footprint reduction and
much faster
thread creation. 

I have it running for about 1 year w/o any noticable
speed penalty.
One of the OpenACS users also tried it and is very
pleased. I think
we can add this as alternate interp initialization scheme.




----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-19 20:30

Message:
Logged In: YES 
user_id=184124

could be a good option, yes

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

[ naviserver-Feature Requests-1166553 ] Add ttrace module

[SourceForge.net <no...@so...>]

Feature Requests item #1166553, was opened at 2005-03-19 17:35
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Zoran Vasiljevic (vasiljevic)
Summary: Add ttrace module

Initial Comment:
The ttrace module is written in Tcl and is part of the
threading
extension and available on SF:

  http://sourceforge.net/projects/ttrace

The basic idea is to use Tcl trace capabilities and
replicate state
of the Tcl interpreter from one to other threads in a
lazy mode.
So, instead of loading everything in one interp and
then using clumsy
introspective script to build the initialization script
for other threads
the ttrace package loads all procedure, namespace (et
al) defs
and stuffs them into a shared array. By overloading the
Tcl [unknown]
command, resources are brought into life as needed. 
This results in drastic memory footprint reduction and
much faster
thread creation. 

I have it running for about 1 year w/o any noticable
speed penalty.
One of the OpenACS users also tried it and is very
pleased. I think
we can add this as alternate interp initialization scheme.




----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1166553&group_id=130646

[ naviserver-Feature Requests-1162223 ] Tcl Callbacks

[SourceForge.net <no...@so...>]

Feature Requests item #1162223, was opened at 2005-03-13 01:22
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Stephen Deasey (sdeasey)
Assigned to: Stephen Deasey (sdeasey)
Summary: Tcl Callbacks

Initial Comment:

The server offers a dual C/Tcl interface.  There are a
number of of places where Tcl code needs to be executed
by the C core, or the core doesn't know whether a
callback is C or Tcl.

I've attached the patch
naviserver-4.1-tclcallbacks.patch.  It extracts the Tcl
callback infrastructure from tclsched.c into
tclcallbacks.c and makes it a public interface.  Here
it is:

Ns_TclNewCallback()
Ns_TclNewCallbackObj()
Ns_TclEvalCallback()
Ns_TclCallbackProc()
Ns_TclFreeCallback()
Ns_TclCallbackArgProc()

It provides a standard interface for Tcl callbacks
throughout the server, handling interp allocation,
introspection, evaluation and cleanup.

Also, tclcallbacks.c is somewhat analogous to
callbacks.c, and so the generic ns_atexit etc.
callbacks which were in tclsched.c have been moved to
the tclcallbacks file.

The first set of code to be converted to the new
interface is tclsched.c itself, and I've converted to
using Ns_ParseObjv() while I was in there.  File size
has dropped ~250 lines.


One thing I'm wondering about is whether the
Ns_TclEvalCallback() routine should take extra args to
send to the Tcl proc.  There seems to be no standard
scheme for ordering the args so there still needs to be
a way to manually construct the command, which is why
the members of the Ns_TclCallback struct are public. 
But for this very reason, it might be worth providing a
standard way to do this so all future callbacks are
consistent.


It would be nice in the future to come up with some way
of registereing these callback points such that you
don't have to create an ns_info subcommand for example
manually.  It should be possible for loadable modules
to register their callback points and have them show up
in the right place.



----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 19:54

Message:
Logged In: YES 
user_id=95086

After thinking awhile, hm.... why not pass *variable* 
number of arguments to the callback instead of one? 
So: 
 
  register_callback script arg ?arg ...? 
 
and stuff script all args in the callback structure? 
This is ultimate flexibility, or? 
 
Or even go to the other extent and use no arguments at all. 
After all, script is script; you can evaluate just about any 
proper Tcl string. The caller routine may or may not supply 
additional arguments as it usually would (example [fcopy]). 
Hm? Or am I missing the point? 
 
 
 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 15:16

Message:
Logged In: YES 
user_id=184124

Yes, let's start slowly, we know the direction we are going,
let 's do it step by step starting with simple things. We
can have generic Tcl callbacks already and all new callbacks
will use it and we will see how flexible this mechanism is.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-17 12:28

Message:
Logged In: YES 
user_id=87254


Yes, I forgot to mention that Tcl callbacks are a little
more general than those for C in callbacks.c.  Those take a
single arg which is the context you registered with the
proc, and return void.   Tcl callbacks on the other hand can
return a result which at the C level gets appended to a
dstring if you pass that in.

The members of the Ns_TclCallback structure are public so
that you can construct a command and args in any format. 
This is the question I was asking originally: should there
be a default recomended way to do this?

So for example Ns_TclEvalCallback() might take varargs
cstrings, which if passed would be added to the tcl command
to execute:

  command ?cbarg1? ?cbarg2? ?context?

Where cbarg2 and cbarg2 are specific to the type of callback
and context is what was registered with the proc by the Tcl
programmer.

Unfortunately that wouldn't cover all the bases as some of
the existing APIs do weird things like place callback
specific args after the context or if no context was passed,
always append an empty string.  For new callbacks, something
standard like the above scheme should be used.


If you look at the patch for virtual hosting you'll see that
two new Tcl callbacks were created and they contain simillar
boiler plate.  This makes me think that maybe there should
be something like an [ns_addcallback serverroot myrootproc
mycontext] interface.

It's a little more complicated than that because some
callbacks are singular, like the serverroot proc, and some
are plural, like filters.  So maybe the thing to do is get
some experience with this simple version and then expand it
later...





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-13 01:46

Message:
Logged In: YES 
user_id=184124

It looks good, unified callback interface would be better
but Tcl filters. traces, requestprocs uses different
arguments/parameters and currently the only way to get info
about them to use ns_info interface and Ns_ArgProc
interface. Tcl_Callback are for simple proc/args callbacks,
filters is special case, see if you can convert them.
But i am for this change.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-13 01:30

Message:
Logged In: YES 
user_id=87254


I've attached a patch which converts the tclrequest.c file
to use the new callback API as an example.  It also uses the
Ns_ParseObjv() API, and I've removed support for the legacy
conn variable.

It removes ~200 lines and all utility functions.  The
callback API provides default Ns_ArgProc introspection, but
this will be much nicer when combined with Vlad's RFE
1161597  :-)



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1162223&group_id=130646

[ naviserver-Feature Requests-1151137 ] New driver API and Udp module

[SourceForge.net <no...@so...>]

Feature Requests item #1151137, was opened at 2005-02-24 16:01
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: New driver API and Udp module

Initial Comment:
Hi guys,

Attached is minor driver extensions which do not change
existing drivers  but add new functionality. There are
some cosmetic changes, like moving some fields in the
Ns_Sock/Ns_Driver structres so they can be accessed
publically and made some private functions public but
functionality preserved as before.

I included udp driver as an example of new API, and
also added ns_sha1 command in the tclmisc.c, it is just
one command and it is uses practically everywhere.

-------------------------------------------------------------------------
To test udp driver i use new ns_udp command:

ossweb:nscp 8> ns_udp send 127.0.0.1 5060 "GET /
HTTP/1.0\n\n"

HTTP/1.0 200 OK
MIME-Version: 1.0
Date: Thu, 24 Feb 2005 05:39:50 GMT
Server: NaviServer/4.0.10
Content-Type: text/html; charset=iso-8859-1
Content-Length: 661
Connection: close

<HEAD><TITLE>Seryakov's Family Intranet</TITLE></HEAD>
.....


----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 18:17

Message:
Logged In: YES 
user_id=184124

>I'm happy to see any of the core changed to add this
>support, but it needs to be justified.  I don't want to
>support multiple different ways of doing the same thing.

No, it is not the same thing. To build different drivers,
API and tools should be flexible enough, if we have only one
way to add a driver, so only drivers which fit will be(if)
added, and if something usefull will come up, will we need
to change the core again to fit different mehtod?

Basically you offered the same thing that i am talking
about: if we supply new DriverThread for UDP driver, the
only common is how to queue it, public NsQueueConn. The rest
is driver/thread specific. Currently, i can live with
specific thread proc using sock callbacks, but when driver
threads will be available, i can switch to it, but still
once i read data from UDP, i need to build Sock/Conn and
queue it. This is what i am offering, make NsQueueConn
public plus/mines changing some other function but just to
only support public NsQueueConn. IT will not conflict with
future global driver changes.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 14:28

Message:
Logged In: YES 
user_id=184124

One good for having NsQueueConn public is to be able to
submit connection threads from memory, without actual
protocol/driver connection. I have data to spawn connection
thread for and i just submit it. This way my server uses
thread/connection pooling without implementing my own thread
management(for example i do this now in my monitoring, but i
could use conn threads instead with better control over the
queue).




----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 14:23

Message:
Logged In: YES 
user_id=184124

Yes, it will work in the long run. Currently i do not have
any problems with having only one DriverThread but having
different driver threads with specific ThreadProc would be
nice. It is just last time i tried new 4.1 driver, i could
not work with it using Mozilla, somehow all connections got
stale and mozilla was excepting more and more data from the
server while server thought it sent everything already. My
point: it needs more testing and even AOL did not start
testing it.

In the short run i am fine with ability to use
sockcallbacks, i just need to be able to submit Sock to the
ConnQueue, couple of small functions i added. If we can
allow it for now and wait till 4.1 driver will be stable i
would be okay.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 13:03

Message:
Logged In: YES 
user_id=87254


I'm not against making NsQueueConn() public per-se, I am
against publishing it as a work-around to creating a good
API for pluggable protocols.

I'm keen to see great support for text based and binary
protocols, packet based and stream based protocols,
implementations which want to use a listen thread and
multiple worker threads, and those which want to do all work
in a single thread.

I'm happy to see any of the core changed to add this
support, but it needs to be justified.  I don't want to
support multiple different ways of doing the same thing.

Maybe we should backport the changes in 4.1 to our tree
before we start hacking into it?  We could leave out the adp
changes etc. as I believe someone said they were broken.  We
should make a point release before doing this.  What do you
think?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 12:46

Message:
Logged In: YES 
user_id=87254


Sorry, I don't have a patch to show how UDP/Unix sockets
could be integrated into nssock.  The 4.1 code base would
seem to be a better starting point as it creates a new
thread for each loaded comm driver.

Here's how it could work:

Unix domain socket support is easy.  Use the same syntax as
for the binder for the address param of the driver module in
the config file.  Call the correct socket listen routine
depending on type within NsStartDrivers().  The nssock
module should require no changes to work with this socket type.

UDP sockets are harder.  The 4.1 code base has changed
NsStartDrivers() to spin off a thread for each loaded comm
module.  As the requirements for stream and packet sockets
are different, a different routine would be passed to
Ns_ThreadCreate() for each type.

I would rename DriverThread() StreamThread() and add a new
routine, PacketThread().  Much of the contents of
DriverThread() can be junked for handling packet based
protocols, but not all should.  The new QueueWait callbacks
and pre-queue filters are just as applicable here as for
stream based protocols.

The advantage of this is that there is a clear distinction
between network transport, protocol, and application. 
Network transports such as SSL/TLS, SSH, etc. can be swapped
in below different protocol implementations.  Many
applications can be built on a protocol implementation if it
provides the standard hooks (registered proc, filters,
QueueWait callbacks, pre-queue filteres...).

Is this enough to give you an idea?  It seems feasable to
me, what do you think?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 04:31

Message:
Logged In: YES 
user_id=184124

Do we have any conclusions about this? Soon i will have to
develop some application with support for SIP over UDP, i
would like to do it on Naviserver and not on my hacked AS.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 16:44

Message:
Logged In: YES 
user_id=184124

And again, i have nothing against it, i said it before, what
i am offering is a littl ebit different, for network based
protocols driver stuff is kay, i just want to be able to
submit conn threads from within the server, not through the
main driver.
If i have already request data, through shared memory, file
or other way, i want to submit wroker thread directly, not
by connecting to myself and submitting data using HTTP or
other protocol i will have to implement just to queue te
request. This is for other applications, not HTTP/Web based,
more like messaging/authentication servers and alike.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 14:41

Message:
Logged In: YES 
user_id=184124

Can you provide what you have to modify in driver/nssock, i
think that will not be easy and will require too many
modifications.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-04 10:08

Message:
Logged In: YES 
user_id=87254


What do you think of my suggestion that the nssock module
and nsd/driver.c could be easily modified to handle udp and
unix domain sockets when specified in the config file?

A change like that combined with the work you've already
done to the binder would mean no development is required to
handle all socket types, and would be transparent to the
upper layers.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-26 17:10

Message:
Logged In: YES 
user_id=184124

Stephan,
I am suggesting a compromise, let's put ParseProc and
DriverSockQueu patches, they can coexists easily but will
offer two way to add new drivers. Once we have new drivers
working we will see what can be modified/adjusted, for now
we do not have anything except my several drivers i use in
the production and i use my own loop for smtp driver to keep
processing and state machine in one place.
Again, i have nothing against your method, i used to do
similar in my previous versions but now i need to be
completely free to implement aditional driver and i think we
should have this ability.

Zoran, what do you think?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 22:57

Message:
Logged In: YES 
user_id=184124

Try to implement HTTP over UDP as i provided in the patch,
you will need to modify core driver again, this is one example.

Next, if i want to implement small driver that do not
interact with web part of aolserver, no filters, it just
need to do one small thing, it should be fast but will
handle very many requests. I do not need filters, callbacks,
i need my own main loop. 
I do not understand why you are insisting that all drivers
should go through http driver's main loop. If i implement
dns server and web interface to it, why they should compete
in the same driver's main loop and go through registered
filters if they share only backend database structure?

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 22:22

Message:
Logged In: YES 
user_id=87254


You don't have to use registered filters or procs.  It's an
option for protocol implementations where the initial accept
and parsing take place in the driver thread, and processing
takes place in a conn thread.

Once you submit to the queue however, either a filter, a
registered proc, or a registered proxy proc must handle the
request.  It makes no difference whether you submit
explicitly with Ns_DriverSockQueue or implicitly after your
Ns_ParseProc completes.

How much of the request structure you fill in is up to you,
it's all optional.  If you choose not to use it there is no
overhead.  It doesn't have to be a text based protocol for
this to make sense.  You might fill in the request structure
if you wanted to enable people to handle different request
types via C or Tcl.  Or, you could store binary data in
Ns_Cls storage and access it from registered procs etc.

In the case where you want to handle connection
accept/parse/reponse processing in the same thread, nothing
needs to be added.  You can do that today with
Ns_SockCallback (or by placing all your code in an
Ns_ParseProc and returning the response from there)

Can you be more specific about the overhead of using an
Ns_ParseProc?  Exactly what memory is allocated, what code
is run that should not..?

What specifically can you not do with the Ns_ParseProc
interface that you need the Ns_Driver* routines for?  I'm
looking for concrete examples.

As far as I can tell, with Ns_ParseProc you write less code,
get more options, and it takes advantage of infrastructure
to give you more speed.  I must be missing something, but
you'll have to explain it to me in more detail because I
don't get it.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 20:43

Message:
Logged In: YES 
user_id=184124

Okay, the biggest problem with your way i see that i am
enforced to go through http driver, even with my parse proc,
i need to know how it works to implement my driver so
correct hooks/filters will be called. If i want skip some
parts i am not able to do it, http driver works only one way. 
Another issue is if something will not fit into current
driver, additional hook needs to be introduced and core
driver needs to be modified again.

If my driver fits http-like/text mode paradigm, reusing http
driver is the easiest way, but if i need something specific,
all extra step to mimic the connection as http request, so 
all other parts of the http driver will not fail is just
unnecesssary extra efforts.

And i you mentioned, using Ns_DriverSockQueue is low level
function and requires from the developer full attention for
bulding the driver, but if this is what i want, why not. I
can spawn my own thread if some callbacks are long running
Tcl scripts and queue connection from my own loop. If i omly
want to received packets, decode them and submit connection,
i do not need all http driver infrusturcture, why to enforce
using it?

So we can have at least 3 ways for supporting multiprotocols:
- standard callbacks, completely different threads, no
connection pooling (Exists now)

- Ns_DriverSockQueue, new thread or callback thread, reusing
connection pooling. All filters/traces are reused as well

- driver Parse proc in the http driver to reuse drivers
thread and connection pooling with all filters/traces

All 3 methiods can co-exists and do not interfere with each
other, that's my point, and all they add just couple extra
functions. I would implement all methods, i have 3 drivers i
need to run: smtp, dns, snmp, can add imap in the future

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 20:31

Message:
Logged In: YES 
user_id=87254


"Each way has its own drawbacks..."

That's what I'm asking.  What are the drawbacks for each
approach we've come up with so far?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 15:23

Message:
Logged In: YES 
user_id=184124

I am not arguing with you about your ParseProc driver
extension, it is very usefull and makes driver more flexible.

I just want to add couple of new API functions that will allow
developer to queue connections from any place, that's it.
How drivers will be written and how developer will decide to
handle it is up to each particular developer. But nobody
will be locked up in only one way of doing it.

Each way has its own drawbacks so we have multiple choices
and developer will decide for itself how it should be.

Both our ways do not mutually exclusive.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 08:28

Message:
Logged In: YES 
user_id=87254


Ah I see you're right, an extra thread is not created for
each incoming request.  'udpThread' is perhaps not the best
name for the socket callback though... :-)

I still don't see the advantage of using Ns_SockCalback.  A
single thread is created by the server to handle all
registered callbacks, including those from ns_sockcallback.
 At runtime, while this single thread is running tcl code to
handle one callback, your dirver callback cannot run, no new
connections will be accepted, etc.  Why is it better to use
the Ns_SockCallback thread rather than a driver thread?

Maybe I'm reading this wrong, but how do you handle the case
where the request arrives as more than one packet? 
Ns_DriverSockRead() is called from the 'udpThread' socket
callback.  The only return value checked is NS_OK, but
couldn't this also be SOCK_MORE?

How would you handle things like keepalive?  I think you'd
have to reimplement that in the Ns_SockCallback thread.

Re the proxy stuff, Ns_RegisterRequest() and
Ns_RegisterProxyRequest() seem very simillar.  With
Ns_RegisterRequest(), filters are run and you get the choice
of using C or Tcl.  Ns_RegisterProxyRequest() offers no
advantage that I can see -- you still need a complete
Request structure, even if you just ignore it.


The comm driver initialization should probably be changed to
automatically handle unix domain sockets.  You add /foo
rather that 127.0.0.1 in the config file and it knows to
create the correct type of socket.  nssock and nsopenssl
wouldn't have to be modified at all.

UDP is different.  There are two types of protocols: single
packet, such as DNS or RADIUS; multi packet, as used in some
streaming media and p2p protocols.

I think for the single packet case we again might want to
modify the driver code to automatically handle it.  No
read-ahead is necessary, there's only one packet per
request, so it's placed in the request buffer and passed on
to the next stage, which is parsing.  Here, a return code of
SOCK_MORE would be illegal.

Every multi-packet UDP protocol will require custom
framing/sequencing and the developer will have to create a
new socket driver.

Taking RADIUS as an example, which is a single packet UDP
protocol, you'd create a very simple Ns_ParseProc who's only
job is to check that the number of bytes specified in the
header actually arrived, and return SOCK_READY.  A default
request structure is created for you so the only other thing
you have to do is Ns_RegisterRequest() for the '/' URL. 
Within your request proc, call Ns_ConnContent() and parse
the buffer.

Now, you do have a number of other options to make this more
flexible.  You could parse the request in your Ns_ParseProc
and then fill out the request structure.  e.g. the different
RADIUS message types could be expressed as HTTP verbs.  This
buys you flexibility.  Now you can Ns_RegisterRequest() a
different routine for each RADIUS message type, and someone
reusing your code can override your default implementation.
 You can also handle some message types in C, and others in Tcl.

You might decide to put some useful information about the
RADIUS request in the URL line.  Now you get logging for free.

You might decide to parse the key/value pairs from the
RADIUS request into query vars or headers.  Now you don't
have to write a bunch of RADIUS specific Tcl commands to
access the data, should you want to handle that from Tcl. 
ns_queryget or ns_conn headers  will work.

How far you want to go is up to you, the writer of the
protocol module.  But at the low end all you need is an
Ns_ParseProc and usually a single registered request proc,
which is a lot less code to write than the Ns_Driver* stuff,
I think.

I'm sure it's not perfect!  So what's wrong with it? Why
would it suck to write a RADIUS protocol module as I've just
described, to take one example?





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:51

Message:
Logged In: YES 
user_id=184124

I think we need to collect all solutions and then see where
are going, i am holding to port all my old/running modules
because i do not know how naviserver will handle foreign
protocols.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:48

Message:
Logged In: YES 
user_id=184124

yes, using proxy can solve immediate requirements without
hacking NsConnProc by adding hooks to call driver specific C
functions.

If i need my smtp server and main loop is in C, i need
somehow call it
in the connection thread. Using registered proxy function i
can do it now, i do not need filter/traces. This is for
completely new modules.

I can implement main loop in the module as Tcl command and
then call it in the connection filter, it is possible, it
will just require many different parts to be in place and
still filter should be registered as Tcl proc which will
call another driver main loop.

sometimes low level stuff makes things easier and simpler:-)))

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:45

Message:
Logged In: YES 
user_id=184124

In all my installations i need sha module but it wasn't in
the distribution, so i need to download it or repackage
aolserver to include nssha1. That is my point, it is just
one simple function use more often than something like
ns_jpegsize or so.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:42

Message:
Logged In: YES 
user_id=87254


Well I don't know about the wisdom of adding sha1 to the
core at this stage  :-)

I see your point though, encryption and hashing functions
are almost universally required for systems/server work. 
Maybe we need to consider adding a new encryption module to
the core distribution.  Like nsdb it would export a C API
via libnscypher.so (or whatever) as well as the Tcl module
nscypher.so.

Times have changed and things like the openssl libraries are
common on all platforms so it's not the big deal it used to
be to add such dependiencies.

Such a module should make implementing SSL in nsopenssl easier.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:41

Message:
Logged In: YES 
user_id=184124

No, i do not spawn new thread, i use callback feature of the
server, when socket is ready, server calls provided callback
ad that callback just submits the socket to connection pool.
if pool is full, Ns_DriverSockQueue will return NS_TIMEOUT,
you can retry.
Yes, it is low-level, for high level, HTTP driver provides a
lot of functionality, it could be extended like you did, but
still it is HTTP driver hacked. If i want completely new
driver, like RADIUS server, http driver will not help me, i
need low level stuff, and it is there already.

To reuse resource limiting, i added Ns_DriverSockQueue
function, so
new conections can be queue instead of creating new threads.

I do not have anything against your patch for extending
current driver, it is very usefull. i am offering new API
for low level drivers.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:36

Message:
Logged In: YES 
user_id=87254


The proxy stuff is too late in the cycle to do much.  The
request has to be fully parsed by then (for read-ahead).

If a proxy function is available, then all filters, the auth
phase, registered procs, cleanup procs etc. are bypassed.  A
lot of that stuff can be very useful for non-HTTP protocols.

I don't think handling stuff via a proxy function buys you much.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:32

Message:
Logged In: YES 
user_id=87254


I don't understand what you're trying to acheive here (well,
apart from multi-protocol... :-).  The newly exposed
Ns_Driver* entry points are quite low level, and so the
implementor of the new protocol is left to do a lot of the
heavy lifting.

For example, the way I read it you have to create your own
listen socket and register a callback.  Every time a new
reaquest comes in a thread is spawned to handle it.  From
that thread you then submit the parsed request to one of the
conn threads.

Excessive thread creation and message passing between
threads is not going to perform well.  And it seems you have
to write more code than e.g. the example POP3 driver I
posted some time ago.

You're also not taking advantage of the other facilities
that the server offers.  What happens if 1000 connections
arrive, do you spawn 1000 threads?  You could of course code
up some limit checks, but this already exists.  What if a
client sends you a continuous stream of data, 2GB...  etc.

By using the driver hooks to provide the new protocol
parser, you deny yourself the opportunity to use something
like the nsopenssl module.  This should work just fine for
protocols like SMTP, IMAP, POP3 and probably others.


Anyway, I think one of the most carefully coded aspects of
the server is it's attention to resource usage.  That goes
for IO, context switching, memory, etc.  It's espescialy
nice that most of the time you're not even aware that all
this work is being done for you.  I'd like new protocol
drivers to be able to transparently take advantage of that.

Could you take a look at my old POP3 demo driver?  It's the
attachment nspopd-0.3.tar.bz2 over here:

http://sourceforge.net/tracker/index.php?func=detail&aid=973010&group_id=3152&atid=353152

It's not obvious that anything interesting is going on, so
it's not much to look at.  But actually, conn socket
read-ahead is happening eficiently in one thread with async
IO, the conn threads are treated as a precious resource
(heavy-weight Tcl interps) and are allocated at the last
minute, there's an easy API in C and Tcl to implement the
actual reading of data from the INBOX (could be from the
file system, db etc.).

You've got a lot of experience writing servers, what do you
think is wrong with this model?  What can it not do, or what
could it do better?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 21:29

Message:
Logged In: YES 
user_id=184124

Another thing, once we can submit connections from any
place, no need to build any drivers, even in C, i can
register new proxy proc and set protocol field in my
request, so when submitted, connection will run registered
proxy proc.
for example: 
 in my smtp driver/module, i create driver, register proxy
for smtp: protocol, register callback for the socket. Once
connection accepted, in my module i submit that connection
to the queue with request-protocl set to smtp:.
queue.c will call my proxy handler, which is C function.

No need to add anything else. This way even standard
aolserver can be extended without touching precious http
driver thread. Sorry for sarcasm.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

[ naviserver-Feature Requests-1151137 ] New driver API and Udp module

[SourceForge.net <no...@so...>]

Feature Requests item #1151137, was opened at 2005-02-24 16:01
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: New driver API and Udp module

Initial Comment:
Hi guys,

Attached is minor driver extensions which do not change
existing drivers  but add new functionality. There are
some cosmetic changes, like moving some fields in the
Ns_Sock/Ns_Driver structres so they can be accessed
publically and made some private functions public but
functionality preserved as before.

I included udp driver as an example of new API, and
also added ns_sha1 command in the tclmisc.c, it is just
one command and it is uses practically everywhere.

-------------------------------------------------------------------------
To test udp driver i use new ns_udp command:

ossweb:nscp 8> ns_udp send 127.0.0.1 5060 "GET /
HTTP/1.0\n\n"

HTTP/1.0 200 OK
MIME-Version: 1.0
Date: Thu, 24 Feb 2005 05:39:50 GMT
Server: NaviServer/4.0.10
Content-Type: text/html; charset=iso-8859-1
Content-Length: 661
Connection: close

<HEAD><TITLE>Seryakov's Family Intranet</TITLE></HEAD>
.....


----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 14:28

Message:
Logged In: YES 
user_id=184124

One good for having NsQueueConn public is to be able to
submit connection threads from memory, without actual
protocol/driver connection. I have data to spawn connection
thread for and i just submit it. This way my server uses
thread/connection pooling without implementing my own thread
management(for example i do this now in my monitoring, but i
could use conn threads instead with better control over the
queue).




----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 14:23

Message:
Logged In: YES 
user_id=184124

Yes, it will work in the long run. Currently i do not have
any problems with having only one DriverThread but having
different driver threads with specific ThreadProc would be
nice. It is just last time i tried new 4.1 driver, i could
not work with it using Mozilla, somehow all connections got
stale and mozilla was excepting more and more data from the
server while server thought it sent everything already. My
point: it needs more testing and even AOL did not start
testing it.

In the short run i am fine with ability to use
sockcallbacks, i just need to be able to submit Sock to the
ConnQueue, couple of small functions i added. If we can
allow it for now and wait till 4.1 driver will be stable i
would be okay.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 13:03

Message:
Logged In: YES 
user_id=87254


I'm not against making NsQueueConn() public per-se, I am
against publishing it as a work-around to creating a good
API for pluggable protocols.

I'm keen to see great support for text based and binary
protocols, packet based and stream based protocols,
implementations which want to use a listen thread and
multiple worker threads, and those which want to do all work
in a single thread.

I'm happy to see any of the core changed to add this
support, but it needs to be justified.  I don't want to
support multiple different ways of doing the same thing.

Maybe we should backport the changes in 4.1 to our tree
before we start hacking into it?  We could leave out the adp
changes etc. as I believe someone said they were broken.  We
should make a point release before doing this.  What do you
think?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 12:46

Message:
Logged In: YES 
user_id=87254


Sorry, I don't have a patch to show how UDP/Unix sockets
could be integrated into nssock.  The 4.1 code base would
seem to be a better starting point as it creates a new
thread for each loaded comm driver.

Here's how it could work:

Unix domain socket support is easy.  Use the same syntax as
for the binder for the address param of the driver module in
the config file.  Call the correct socket listen routine
depending on type within NsStartDrivers().  The nssock
module should require no changes to work with this socket type.

UDP sockets are harder.  The 4.1 code base has changed
NsStartDrivers() to spin off a thread for each loaded comm
module.  As the requirements for stream and packet sockets
are different, a different routine would be passed to
Ns_ThreadCreate() for each type.

I would rename DriverThread() StreamThread() and add a new
routine, PacketThread().  Much of the contents of
DriverThread() can be junked for handling packet based
protocols, but not all should.  The new QueueWait callbacks
and pre-queue filters are just as applicable here as for
stream based protocols.

The advantage of this is that there is a clear distinction
between network transport, protocol, and application. 
Network transports such as SSL/TLS, SSH, etc. can be swapped
in below different protocol implementations.  Many
applications can be built on a protocol implementation if it
provides the standard hooks (registered proc, filters,
QueueWait callbacks, pre-queue filteres...).

Is this enough to give you an idea?  It seems feasable to
me, what do you think?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 04:31

Message:
Logged In: YES 
user_id=184124

Do we have any conclusions about this? Soon i will have to
develop some application with support for SIP over UDP, i
would like to do it on Naviserver and not on my hacked AS.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 16:44

Message:
Logged In: YES 
user_id=184124

And again, i have nothing against it, i said it before, what
i am offering is a littl ebit different, for network based
protocols driver stuff is kay, i just want to be able to
submit conn threads from within the server, not through the
main driver.
If i have already request data, through shared memory, file
or other way, i want to submit wroker thread directly, not
by connecting to myself and submitting data using HTTP or
other protocol i will have to implement just to queue te
request. This is for other applications, not HTTP/Web based,
more like messaging/authentication servers and alike.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 14:41

Message:
Logged In: YES 
user_id=184124

Can you provide what you have to modify in driver/nssock, i
think that will not be easy and will require too many
modifications.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-04 10:08

Message:
Logged In: YES 
user_id=87254


What do you think of my suggestion that the nssock module
and nsd/driver.c could be easily modified to handle udp and
unix domain sockets when specified in the config file?

A change like that combined with the work you've already
done to the binder would mean no development is required to
handle all socket types, and would be transparent to the
upper layers.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-26 17:10

Message:
Logged In: YES 
user_id=184124

Stephan,
I am suggesting a compromise, let's put ParseProc and
DriverSockQueu patches, they can coexists easily but will
offer two way to add new drivers. Once we have new drivers
working we will see what can be modified/adjusted, for now
we do not have anything except my several drivers i use in
the production and i use my own loop for smtp driver to keep
processing and state machine in one place.
Again, i have nothing against your method, i used to do
similar in my previous versions but now i need to be
completely free to implement aditional driver and i think we
should have this ability.

Zoran, what do you think?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 22:57

Message:
Logged In: YES 
user_id=184124

Try to implement HTTP over UDP as i provided in the patch,
you will need to modify core driver again, this is one example.

Next, if i want to implement small driver that do not
interact with web part of aolserver, no filters, it just
need to do one small thing, it should be fast but will
handle very many requests. I do not need filters, callbacks,
i need my own main loop. 
I do not understand why you are insisting that all drivers
should go through http driver's main loop. If i implement
dns server and web interface to it, why they should compete
in the same driver's main loop and go through registered
filters if they share only backend database structure?

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 22:22

Message:
Logged In: YES 
user_id=87254


You don't have to use registered filters or procs.  It's an
option for protocol implementations where the initial accept
and parsing take place in the driver thread, and processing
takes place in a conn thread.

Once you submit to the queue however, either a filter, a
registered proc, or a registered proxy proc must handle the
request.  It makes no difference whether you submit
explicitly with Ns_DriverSockQueue or implicitly after your
Ns_ParseProc completes.

How much of the request structure you fill in is up to you,
it's all optional.  If you choose not to use it there is no
overhead.  It doesn't have to be a text based protocol for
this to make sense.  You might fill in the request structure
if you wanted to enable people to handle different request
types via C or Tcl.  Or, you could store binary data in
Ns_Cls storage and access it from registered procs etc.

In the case where you want to handle connection
accept/parse/reponse processing in the same thread, nothing
needs to be added.  You can do that today with
Ns_SockCallback (or by placing all your code in an
Ns_ParseProc and returning the response from there)

Can you be more specific about the overhead of using an
Ns_ParseProc?  Exactly what memory is allocated, what code
is run that should not..?

What specifically can you not do with the Ns_ParseProc
interface that you need the Ns_Driver* routines for?  I'm
looking for concrete examples.

As far as I can tell, with Ns_ParseProc you write less code,
get more options, and it takes advantage of infrastructure
to give you more speed.  I must be missing something, but
you'll have to explain it to me in more detail because I
don't get it.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 20:43

Message:
Logged In: YES 
user_id=184124

Okay, the biggest problem with your way i see that i am
enforced to go through http driver, even with my parse proc,
i need to know how it works to implement my driver so
correct hooks/filters will be called. If i want skip some
parts i am not able to do it, http driver works only one way. 
Another issue is if something will not fit into current
driver, additional hook needs to be introduced and core
driver needs to be modified again.

If my driver fits http-like/text mode paradigm, reusing http
driver is the easiest way, but if i need something specific,
all extra step to mimic the connection as http request, so 
all other parts of the http driver will not fail is just
unnecesssary extra efforts.

And i you mentioned, using Ns_DriverSockQueue is low level
function and requires from the developer full attention for
bulding the driver, but if this is what i want, why not. I
can spawn my own thread if some callbacks are long running
Tcl scripts and queue connection from my own loop. If i omly
want to received packets, decode them and submit connection,
i do not need all http driver infrusturcture, why to enforce
using it?

So we can have at least 3 ways for supporting multiprotocols:
- standard callbacks, completely different threads, no
connection pooling (Exists now)

- Ns_DriverSockQueue, new thread or callback thread, reusing
connection pooling. All filters/traces are reused as well

- driver Parse proc in the http driver to reuse drivers
thread and connection pooling with all filters/traces

All 3 methiods can co-exists and do not interfere with each
other, that's my point, and all they add just couple extra
functions. I would implement all methods, i have 3 drivers i
need to run: smtp, dns, snmp, can add imap in the future

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 20:31

Message:
Logged In: YES 
user_id=87254


"Each way has its own drawbacks..."

That's what I'm asking.  What are the drawbacks for each
approach we've come up with so far?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 15:23

Message:
Logged In: YES 
user_id=184124

I am not arguing with you about your ParseProc driver
extension, it is very usefull and makes driver more flexible.

I just want to add couple of new API functions that will allow
developer to queue connections from any place, that's it.
How drivers will be written and how developer will decide to
handle it is up to each particular developer. But nobody
will be locked up in only one way of doing it.

Each way has its own drawbacks so we have multiple choices
and developer will decide for itself how it should be.

Both our ways do not mutually exclusive.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 08:28

Message:
Logged In: YES 
user_id=87254


Ah I see you're right, an extra thread is not created for
each incoming request.  'udpThread' is perhaps not the best
name for the socket callback though... :-)

I still don't see the advantage of using Ns_SockCalback.  A
single thread is created by the server to handle all
registered callbacks, including those from ns_sockcallback.
 At runtime, while this single thread is running tcl code to
handle one callback, your dirver callback cannot run, no new
connections will be accepted, etc.  Why is it better to use
the Ns_SockCallback thread rather than a driver thread?

Maybe I'm reading this wrong, but how do you handle the case
where the request arrives as more than one packet? 
Ns_DriverSockRead() is called from the 'udpThread' socket
callback.  The only return value checked is NS_OK, but
couldn't this also be SOCK_MORE?

How would you handle things like keepalive?  I think you'd
have to reimplement that in the Ns_SockCallback thread.

Re the proxy stuff, Ns_RegisterRequest() and
Ns_RegisterProxyRequest() seem very simillar.  With
Ns_RegisterRequest(), filters are run and you get the choice
of using C or Tcl.  Ns_RegisterProxyRequest() offers no
advantage that I can see -- you still need a complete
Request structure, even if you just ignore it.


The comm driver initialization should probably be changed to
automatically handle unix domain sockets.  You add /foo
rather that 127.0.0.1 in the config file and it knows to
create the correct type of socket.  nssock and nsopenssl
wouldn't have to be modified at all.

UDP is different.  There are two types of protocols: single
packet, such as DNS or RADIUS; multi packet, as used in some
streaming media and p2p protocols.

I think for the single packet case we again might want to
modify the driver code to automatically handle it.  No
read-ahead is necessary, there's only one packet per
request, so it's placed in the request buffer and passed on
to the next stage, which is parsing.  Here, a return code of
SOCK_MORE would be illegal.

Every multi-packet UDP protocol will require custom
framing/sequencing and the developer will have to create a
new socket driver.

Taking RADIUS as an example, which is a single packet UDP
protocol, you'd create a very simple Ns_ParseProc who's only
job is to check that the number of bytes specified in the
header actually arrived, and return SOCK_READY.  A default
request structure is created for you so the only other thing
you have to do is Ns_RegisterRequest() for the '/' URL. 
Within your request proc, call Ns_ConnContent() and parse
the buffer.

Now, you do have a number of other options to make this more
flexible.  You could parse the request in your Ns_ParseProc
and then fill out the request structure.  e.g. the different
RADIUS message types could be expressed as HTTP verbs.  This
buys you flexibility.  Now you can Ns_RegisterRequest() a
different routine for each RADIUS message type, and someone
reusing your code can override your default implementation.
 You can also handle some message types in C, and others in Tcl.

You might decide to put some useful information about the
RADIUS request in the URL line.  Now you get logging for free.

You might decide to parse the key/value pairs from the
RADIUS request into query vars or headers.  Now you don't
have to write a bunch of RADIUS specific Tcl commands to
access the data, should you want to handle that from Tcl. 
ns_queryget or ns_conn headers  will work.

How far you want to go is up to you, the writer of the
protocol module.  But at the low end all you need is an
Ns_ParseProc and usually a single registered request proc,
which is a lot less code to write than the Ns_Driver* stuff,
I think.

I'm sure it's not perfect!  So what's wrong with it? Why
would it suck to write a RADIUS protocol module as I've just
described, to take one example?





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:51

Message:
Logged In: YES 
user_id=184124

I think we need to collect all solutions and then see where
are going, i am holding to port all my old/running modules
because i do not know how naviserver will handle foreign
protocols.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:48

Message:
Logged In: YES 
user_id=184124

yes, using proxy can solve immediate requirements without
hacking NsConnProc by adding hooks to call driver specific C
functions.

If i need my smtp server and main loop is in C, i need
somehow call it
in the connection thread. Using registered proxy function i
can do it now, i do not need filter/traces. This is for
completely new modules.

I can implement main loop in the module as Tcl command and
then call it in the connection filter, it is possible, it
will just require many different parts to be in place and
still filter should be registered as Tcl proc which will
call another driver main loop.

sometimes low level stuff makes things easier and simpler:-)))

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:45

Message:
Logged In: YES 
user_id=184124

In all my installations i need sha module but it wasn't in
the distribution, so i need to download it or repackage
aolserver to include nssha1. That is my point, it is just
one simple function use more often than something like
ns_jpegsize or so.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:42

Message:
Logged In: YES 
user_id=87254


Well I don't know about the wisdom of adding sha1 to the
core at this stage  :-)

I see your point though, encryption and hashing functions
are almost universally required for systems/server work. 
Maybe we need to consider adding a new encryption module to
the core distribution.  Like nsdb it would export a C API
via libnscypher.so (or whatever) as well as the Tcl module
nscypher.so.

Times have changed and things like the openssl libraries are
common on all platforms so it's not the big deal it used to
be to add such dependiencies.

Such a module should make implementing SSL in nsopenssl easier.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:41

Message:
Logged In: YES 
user_id=184124

No, i do not spawn new thread, i use callback feature of the
server, when socket is ready, server calls provided callback
ad that callback just submits the socket to connection pool.
if pool is full, Ns_DriverSockQueue will return NS_TIMEOUT,
you can retry.
Yes, it is low-level, for high level, HTTP driver provides a
lot of functionality, it could be extended like you did, but
still it is HTTP driver hacked. If i want completely new
driver, like RADIUS server, http driver will not help me, i
need low level stuff, and it is there already.

To reuse resource limiting, i added Ns_DriverSockQueue
function, so
new conections can be queue instead of creating new threads.

I do not have anything against your patch for extending
current driver, it is very usefull. i am offering new API
for low level drivers.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:36

Message:
Logged In: YES 
user_id=87254


The proxy stuff is too late in the cycle to do much.  The
request has to be fully parsed by then (for read-ahead).

If a proxy function is available, then all filters, the auth
phase, registered procs, cleanup procs etc. are bypassed.  A
lot of that stuff can be very useful for non-HTTP protocols.

I don't think handling stuff via a proxy function buys you much.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:32

Message:
Logged In: YES 
user_id=87254


I don't understand what you're trying to acheive here (well,
apart from multi-protocol... :-).  The newly exposed
Ns_Driver* entry points are quite low level, and so the
implementor of the new protocol is left to do a lot of the
heavy lifting.

For example, the way I read it you have to create your own
listen socket and register a callback.  Every time a new
reaquest comes in a thread is spawned to handle it.  From
that thread you then submit the parsed request to one of the
conn threads.

Excessive thread creation and message passing between
threads is not going to perform well.  And it seems you have
to write more code than e.g. the example POP3 driver I
posted some time ago.

You're also not taking advantage of the other facilities
that the server offers.  What happens if 1000 connections
arrive, do you spawn 1000 threads?  You could of course code
up some limit checks, but this already exists.  What if a
client sends you a continuous stream of data, 2GB...  etc.

By using the driver hooks to provide the new protocol
parser, you deny yourself the opportunity to use something
like the nsopenssl module.  This should work just fine for
protocols like SMTP, IMAP, POP3 and probably others.


Anyway, I think one of the most carefully coded aspects of
the server is it's attention to resource usage.  That goes
for IO, context switching, memory, etc.  It's espescialy
nice that most of the time you're not even aware that all
this work is being done for you.  I'd like new protocol
drivers to be able to transparently take advantage of that.

Could you take a look at my old POP3 demo driver?  It's the
attachment nspopd-0.3.tar.bz2 over here:

http://sourceforge.net/tracker/index.php?func=detail&aid=973010&group_id=3152&atid=353152

It's not obvious that anything interesting is going on, so
it's not much to look at.  But actually, conn socket
read-ahead is happening eficiently in one thread with async
IO, the conn threads are treated as a precious resource
(heavy-weight Tcl interps) and are allocated at the last
minute, there's an easy API in C and Tcl to implement the
actual reading of data from the INBOX (could be from the
file system, db etc.).

You've got a lot of experience writing servers, what do you
think is wrong with this model?  What can it not do, or what
could it do better?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 21:29

Message:
Logged In: YES 
user_id=184124

Another thing, once we can submit connections from any
place, no need to build any drivers, even in C, i can
register new proxy proc and set protocol field in my
request, so when submitted, connection will run registered
proxy proc.
for example: 
 in my smtp driver/module, i create driver, register proxy
for smtp: protocol, register callback for the socket. Once
connection accepted, in my module i submit that connection
to the queue with request-protocl set to smtp:.
queue.c will call my proxy handler, which is C function.

No need to add anything else. This way even standard
aolserver can be extended without touching precious http
driver thread. Sorry for sarcasm.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

[ naviserver-Feature Requests-1151137 ] New driver API and Udp module

[SourceForge.net <no...@so...>]

Feature Requests item #1151137, was opened at 2005-02-24 16:01
Message generated for change (Comment added) made by seryakov
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: New driver API and Udp module

Initial Comment:
Hi guys,

Attached is minor driver extensions which do not change
existing drivers  but add new functionality. There are
some cosmetic changes, like moving some fields in the
Ns_Sock/Ns_Driver structres so they can be accessed
publically and made some private functions public but
functionality preserved as before.

I included udp driver as an example of new API, and
also added ns_sha1 command in the tclmisc.c, it is just
one command and it is uses practically everywhere.

-------------------------------------------------------------------------
To test udp driver i use new ns_udp command:

ossweb:nscp 8> ns_udp send 127.0.0.1 5060 "GET /
HTTP/1.0\n\n"

HTTP/1.0 200 OK
MIME-Version: 1.0
Date: Thu, 24 Feb 2005 05:39:50 GMT
Server: NaviServer/4.0.10
Content-Type: text/html; charset=iso-8859-1
Content-Length: 661
Connection: close

<HEAD><TITLE>Seryakov's Family Intranet</TITLE></HEAD>
.....


----------------------------------------------------------------------

>Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 14:23

Message:
Logged In: YES 
user_id=184124

Yes, it will work in the long run. Currently i do not have
any problems with having only one DriverThread but having
different driver threads with specific ThreadProc would be
nice. It is just last time i tried new 4.1 driver, i could
not work with it using Mozilla, somehow all connections got
stale and mozilla was excepting more and more data from the
server while server thought it sent everything already. My
point: it needs more testing and even AOL did not start
testing it.

In the short run i am fine with ability to use
sockcallbacks, i just need to be able to submit Sock to the
ConnQueue, couple of small functions i added. If we can
allow it for now and wait till 4.1 driver will be stable i
would be okay.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 13:03

Message:
Logged In: YES 
user_id=87254


I'm not against making NsQueueConn() public per-se, I am
against publishing it as a work-around to creating a good
API for pluggable protocols.

I'm keen to see great support for text based and binary
protocols, packet based and stream based protocols,
implementations which want to use a listen thread and
multiple worker threads, and those which want to do all work
in a single thread.

I'm happy to see any of the core changed to add this
support, but it needs to be justified.  I don't want to
support multiple different ways of doing the same thing.

Maybe we should backport the changes in 4.1 to our tree
before we start hacking into it?  We could leave out the adp
changes etc. as I believe someone said they were broken.  We
should make a point release before doing this.  What do you
think?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 12:46

Message:
Logged In: YES 
user_id=87254


Sorry, I don't have a patch to show how UDP/Unix sockets
could be integrated into nssock.  The 4.1 code base would
seem to be a better starting point as it creates a new
thread for each loaded comm driver.

Here's how it could work:

Unix domain socket support is easy.  Use the same syntax as
for the binder for the address param of the driver module in
the config file.  Call the correct socket listen routine
depending on type within NsStartDrivers().  The nssock
module should require no changes to work with this socket type.

UDP sockets are harder.  The 4.1 code base has changed
NsStartDrivers() to spin off a thread for each loaded comm
module.  As the requirements for stream and packet sockets
are different, a different routine would be passed to
Ns_ThreadCreate() for each type.

I would rename DriverThread() StreamThread() and add a new
routine, PacketThread().  Much of the contents of
DriverThread() can be junked for handling packet based
protocols, but not all should.  The new QueueWait callbacks
and pre-queue filters are just as applicable here as for
stream based protocols.

The advantage of this is that there is a clear distinction
between network transport, protocol, and application. 
Network transports such as SSL/TLS, SSH, etc. can be swapped
in below different protocol implementations.  Many
applications can be built on a protocol implementation if it
provides the standard hooks (registered proc, filters,
QueueWait callbacks, pre-queue filteres...).

Is this enough to give you an idea?  It seems feasable to
me, what do you think?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 04:31

Message:
Logged In: YES 
user_id=184124

Do we have any conclusions about this? Soon i will have to
develop some application with support for SIP over UDP, i
would like to do it on Naviserver and not on my hacked AS.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 16:44

Message:
Logged In: YES 
user_id=184124

And again, i have nothing against it, i said it before, what
i am offering is a littl ebit different, for network based
protocols driver stuff is kay, i just want to be able to
submit conn threads from within the server, not through the
main driver.
If i have already request data, through shared memory, file
or other way, i want to submit wroker thread directly, not
by connecting to myself and submitting data using HTTP or
other protocol i will have to implement just to queue te
request. This is for other applications, not HTTP/Web based,
more like messaging/authentication servers and alike.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 14:41

Message:
Logged In: YES 
user_id=184124

Can you provide what you have to modify in driver/nssock, i
think that will not be easy and will require too many
modifications.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-04 10:08

Message:
Logged In: YES 
user_id=87254


What do you think of my suggestion that the nssock module
and nsd/driver.c could be easily modified to handle udp and
unix domain sockets when specified in the config file?

A change like that combined with the work you've already
done to the binder would mean no development is required to
handle all socket types, and would be transparent to the
upper layers.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-26 17:10

Message:
Logged In: YES 
user_id=184124

Stephan,
I am suggesting a compromise, let's put ParseProc and
DriverSockQueu patches, they can coexists easily but will
offer two way to add new drivers. Once we have new drivers
working we will see what can be modified/adjusted, for now
we do not have anything except my several drivers i use in
the production and i use my own loop for smtp driver to keep
processing and state machine in one place.
Again, i have nothing against your method, i used to do
similar in my previous versions but now i need to be
completely free to implement aditional driver and i think we
should have this ability.

Zoran, what do you think?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 22:57

Message:
Logged In: YES 
user_id=184124

Try to implement HTTP over UDP as i provided in the patch,
you will need to modify core driver again, this is one example.

Next, if i want to implement small driver that do not
interact with web part of aolserver, no filters, it just
need to do one small thing, it should be fast but will
handle very many requests. I do not need filters, callbacks,
i need my own main loop. 
I do not understand why you are insisting that all drivers
should go through http driver's main loop. If i implement
dns server and web interface to it, why they should compete
in the same driver's main loop and go through registered
filters if they share only backend database structure?

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 22:22

Message:
Logged In: YES 
user_id=87254


You don't have to use registered filters or procs.  It's an
option for protocol implementations where the initial accept
and parsing take place in the driver thread, and processing
takes place in a conn thread.

Once you submit to the queue however, either a filter, a
registered proc, or a registered proxy proc must handle the
request.  It makes no difference whether you submit
explicitly with Ns_DriverSockQueue or implicitly after your
Ns_ParseProc completes.

How much of the request structure you fill in is up to you,
it's all optional.  If you choose not to use it there is no
overhead.  It doesn't have to be a text based protocol for
this to make sense.  You might fill in the request structure
if you wanted to enable people to handle different request
types via C or Tcl.  Or, you could store binary data in
Ns_Cls storage and access it from registered procs etc.

In the case where you want to handle connection
accept/parse/reponse processing in the same thread, nothing
needs to be added.  You can do that today with
Ns_SockCallback (or by placing all your code in an
Ns_ParseProc and returning the response from there)

Can you be more specific about the overhead of using an
Ns_ParseProc?  Exactly what memory is allocated, what code
is run that should not..?

What specifically can you not do with the Ns_ParseProc
interface that you need the Ns_Driver* routines for?  I'm
looking for concrete examples.

As far as I can tell, with Ns_ParseProc you write less code,
get more options, and it takes advantage of infrastructure
to give you more speed.  I must be missing something, but
you'll have to explain it to me in more detail because I
don't get it.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 20:43

Message:
Logged In: YES 
user_id=184124

Okay, the biggest problem with your way i see that i am
enforced to go through http driver, even with my parse proc,
i need to know how it works to implement my driver so
correct hooks/filters will be called. If i want skip some
parts i am not able to do it, http driver works only one way. 
Another issue is if something will not fit into current
driver, additional hook needs to be introduced and core
driver needs to be modified again.

If my driver fits http-like/text mode paradigm, reusing http
driver is the easiest way, but if i need something specific,
all extra step to mimic the connection as http request, so 
all other parts of the http driver will not fail is just
unnecesssary extra efforts.

And i you mentioned, using Ns_DriverSockQueue is low level
function and requires from the developer full attention for
bulding the driver, but if this is what i want, why not. I
can spawn my own thread if some callbacks are long running
Tcl scripts and queue connection from my own loop. If i omly
want to received packets, decode them and submit connection,
i do not need all http driver infrusturcture, why to enforce
using it?

So we can have at least 3 ways for supporting multiprotocols:
- standard callbacks, completely different threads, no
connection pooling (Exists now)

- Ns_DriverSockQueue, new thread or callback thread, reusing
connection pooling. All filters/traces are reused as well

- driver Parse proc in the http driver to reuse drivers
thread and connection pooling with all filters/traces

All 3 methiods can co-exists and do not interfere with each
other, that's my point, and all they add just couple extra
functions. I would implement all methods, i have 3 drivers i
need to run: smtp, dns, snmp, can add imap in the future

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 20:31

Message:
Logged In: YES 
user_id=87254


"Each way has its own drawbacks..."

That's what I'm asking.  What are the drawbacks for each
approach we've come up with so far?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 15:23

Message:
Logged In: YES 
user_id=184124

I am not arguing with you about your ParseProc driver
extension, it is very usefull and makes driver more flexible.

I just want to add couple of new API functions that will allow
developer to queue connections from any place, that's it.
How drivers will be written and how developer will decide to
handle it is up to each particular developer. But nobody
will be locked up in only one way of doing it.

Each way has its own drawbacks so we have multiple choices
and developer will decide for itself how it should be.

Both our ways do not mutually exclusive.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 08:28

Message:
Logged In: YES 
user_id=87254


Ah I see you're right, an extra thread is not created for
each incoming request.  'udpThread' is perhaps not the best
name for the socket callback though... :-)

I still don't see the advantage of using Ns_SockCalback.  A
single thread is created by the server to handle all
registered callbacks, including those from ns_sockcallback.
 At runtime, while this single thread is running tcl code to
handle one callback, your dirver callback cannot run, no new
connections will be accepted, etc.  Why is it better to use
the Ns_SockCallback thread rather than a driver thread?

Maybe I'm reading this wrong, but how do you handle the case
where the request arrives as more than one packet? 
Ns_DriverSockRead() is called from the 'udpThread' socket
callback.  The only return value checked is NS_OK, but
couldn't this also be SOCK_MORE?

How would you handle things like keepalive?  I think you'd
have to reimplement that in the Ns_SockCallback thread.

Re the proxy stuff, Ns_RegisterRequest() and
Ns_RegisterProxyRequest() seem very simillar.  With
Ns_RegisterRequest(), filters are run and you get the choice
of using C or Tcl.  Ns_RegisterProxyRequest() offers no
advantage that I can see -- you still need a complete
Request structure, even if you just ignore it.


The comm driver initialization should probably be changed to
automatically handle unix domain sockets.  You add /foo
rather that 127.0.0.1 in the config file and it knows to
create the correct type of socket.  nssock and nsopenssl
wouldn't have to be modified at all.

UDP is different.  There are two types of protocols: single
packet, such as DNS or RADIUS; multi packet, as used in some
streaming media and p2p protocols.

I think for the single packet case we again might want to
modify the driver code to automatically handle it.  No
read-ahead is necessary, there's only one packet per
request, so it's placed in the request buffer and passed on
to the next stage, which is parsing.  Here, a return code of
SOCK_MORE would be illegal.

Every multi-packet UDP protocol will require custom
framing/sequencing and the developer will have to create a
new socket driver.

Taking RADIUS as an example, which is a single packet UDP
protocol, you'd create a very simple Ns_ParseProc who's only
job is to check that the number of bytes specified in the
header actually arrived, and return SOCK_READY.  A default
request structure is created for you so the only other thing
you have to do is Ns_RegisterRequest() for the '/' URL. 
Within your request proc, call Ns_ConnContent() and parse
the buffer.

Now, you do have a number of other options to make this more
flexible.  You could parse the request in your Ns_ParseProc
and then fill out the request structure.  e.g. the different
RADIUS message types could be expressed as HTTP verbs.  This
buys you flexibility.  Now you can Ns_RegisterRequest() a
different routine for each RADIUS message type, and someone
reusing your code can override your default implementation.
 You can also handle some message types in C, and others in Tcl.

You might decide to put some useful information about the
RADIUS request in the URL line.  Now you get logging for free.

You might decide to parse the key/value pairs from the
RADIUS request into query vars or headers.  Now you don't
have to write a bunch of RADIUS specific Tcl commands to
access the data, should you want to handle that from Tcl. 
ns_queryget or ns_conn headers  will work.

How far you want to go is up to you, the writer of the
protocol module.  But at the low end all you need is an
Ns_ParseProc and usually a single registered request proc,
which is a lot less code to write than the Ns_Driver* stuff,
I think.

I'm sure it's not perfect!  So what's wrong with it? Why
would it suck to write a RADIUS protocol module as I've just
described, to take one example?





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:51

Message:
Logged In: YES 
user_id=184124

I think we need to collect all solutions and then see where
are going, i am holding to port all my old/running modules
because i do not know how naviserver will handle foreign
protocols.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:48

Message:
Logged In: YES 
user_id=184124

yes, using proxy can solve immediate requirements without
hacking NsConnProc by adding hooks to call driver specific C
functions.

If i need my smtp server and main loop is in C, i need
somehow call it
in the connection thread. Using registered proxy function i
can do it now, i do not need filter/traces. This is for
completely new modules.

I can implement main loop in the module as Tcl command and
then call it in the connection filter, it is possible, it
will just require many different parts to be in place and
still filter should be registered as Tcl proc which will
call another driver main loop.

sometimes low level stuff makes things easier and simpler:-)))

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:45

Message:
Logged In: YES 
user_id=184124

In all my installations i need sha module but it wasn't in
the distribution, so i need to download it or repackage
aolserver to include nssha1. That is my point, it is just
one simple function use more often than something like
ns_jpegsize or so.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:42

Message:
Logged In: YES 
user_id=87254


Well I don't know about the wisdom of adding sha1 to the
core at this stage  :-)

I see your point though, encryption and hashing functions
are almost universally required for systems/server work. 
Maybe we need to consider adding a new encryption module to
the core distribution.  Like nsdb it would export a C API
via libnscypher.so (or whatever) as well as the Tcl module
nscypher.so.

Times have changed and things like the openssl libraries are
common on all platforms so it's not the big deal it used to
be to add such dependiencies.

Such a module should make implementing SSL in nsopenssl easier.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 05:41

Message:
Logged In: YES 
user_id=184124

No, i do not spawn new thread, i use callback feature of the
server, when socket is ready, server calls provided callback
ad that callback just submits the socket to connection pool.
if pool is full, Ns_DriverSockQueue will return NS_TIMEOUT,
you can retry.
Yes, it is low-level, for high level, HTTP driver provides a
lot of functionality, it could be extended like you did, but
still it is HTTP driver hacked. If i want completely new
driver, like RADIUS server, http driver will not help me, i
need low level stuff, and it is there already.

To reuse resource limiting, i added Ns_DriverSockQueue
function, so
new conections can be queue instead of creating new threads.

I do not have anything against your patch for extending
current driver, it is very usefull. i am offering new API
for low level drivers.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:36

Message:
Logged In: YES 
user_id=87254


The proxy stuff is too late in the cycle to do much.  The
request has to be fully parsed by then (for read-ahead).

If a proxy function is available, then all filters, the auth
phase, registered procs, cleanup procs etc. are bypassed.  A
lot of that stuff can be very useful for non-HTTP protocols.

I don't think handling stuff via a proxy function buys you much.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 05:32

Message:
Logged In: YES 
user_id=87254


I don't understand what you're trying to acheive here (well,
apart from multi-protocol... :-).  The newly exposed
Ns_Driver* entry points are quite low level, and so the
implementor of the new protocol is left to do a lot of the
heavy lifting.

For example, the way I read it you have to create your own
listen socket and register a callback.  Every time a new
reaquest comes in a thread is spawned to handle it.  From
that thread you then submit the parsed request to one of the
conn threads.

Excessive thread creation and message passing between
threads is not going to perform well.  And it seems you have
to write more code than e.g. the example POP3 driver I
posted some time ago.

You're also not taking advantage of the other facilities
that the server offers.  What happens if 1000 connections
arrive, do you spawn 1000 threads?  You could of course code
up some limit checks, but this already exists.  What if a
client sends you a continuous stream of data, 2GB...  etc.

By using the driver hooks to provide the new protocol
parser, you deny yourself the opportunity to use something
like the nsopenssl module.  This should work just fine for
protocols like SMTP, IMAP, POP3 and probably others.


Anyway, I think one of the most carefully coded aspects of
the server is it's attention to resource usage.  That goes
for IO, context switching, memory, etc.  It's espescialy
nice that most of the time you're not even aware that all
this work is being done for you.  I'd like new protocol
drivers to be able to transparently take advantage of that.

Could you take a look at my old POP3 demo driver?  It's the
attachment nspopd-0.3.tar.bz2 over here:

http://sourceforge.net/tracker/index.php?func=detail&aid=973010&group_id=3152&atid=353152

It's not obvious that anything interesting is going on, so
it's not much to look at.  But actually, conn socket
read-ahead is happening eficiently in one thread with async
IO, the conn threads are treated as a precious resource
(heavy-weight Tcl interps) and are allocated at the last
minute, there's an easy API in C and Tcl to implement the
actual reading of data from the INBOX (could be from the
file system, db etc.).

You've got a lot of experience writing servers, what do you
think is wrong with this model?  What can it not do, or what
could it do better?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 21:29

Message:
Logged In: YES 
user_id=184124

Another thing, once we can submit connections from any
place, no need to build any drivers, even in C, i can
register new proxy proc and set protocol field in my
request, so when submitted, connection will run registered
proxy proc.
for example: 
 in my smtp driver/module, i create driver, register proxy
for smtp: protocol, register callback for the socket. Once
connection accepted, in my module i submit that connection
to the queue with request-protocl set to smtp:.
queue.c will call my proxy handler, which is C function.

No need to add anything else. This way even standard
aolserver can be extended without touching precious http
driver thread. Sorry for sarcasm.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

[ naviserver-Feature Requests-1151137 ] New driver API and Udp module

[SourceForge.net <no...@so...>]

Feature Requests item #1151137, was opened at 2005-02-24 09:01
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: New driver API and Udp module

Initial Comment:
Hi guys,

Attached is minor driver extensions which do not change
existing drivers  but add new functionality. There are
some cosmetic changes, like moving some fields in the
Ns_Sock/Ns_Driver structres so they can be accessed
publically and made some private functions public but
functionality preserved as before.

I included udp driver as an example of new API, and
also added ns_sha1 command in the tclmisc.c, it is just
one command and it is uses practically everywhere.

-------------------------------------------------------------------------
To test udp driver i use new ns_udp command:

ossweb:nscp 8> ns_udp send 127.0.0.1 5060 "GET /
HTTP/1.0\n\n"

HTTP/1.0 200 OK
MIME-Version: 1.0
Date: Thu, 24 Feb 2005 05:39:50 GMT
Server: NaviServer/4.0.10
Content-Type: text/html; charset=iso-8859-1
Content-Length: 661
Connection: close

<HEAD><TITLE>Seryakov's Family Intranet</TITLE></HEAD>
.....


----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 06:03

Message:
Logged In: YES 
user_id=87254


I'm not against making NsQueueConn() public per-se, I am
against publishing it as a work-around to creating a good
API for pluggable protocols.

I'm keen to see great support for text based and binary
protocols, packet based and stream based protocols,
implementations which want to use a listen thread and
multiple worker threads, and those which want to do all work
in a single thread.

I'm happy to see any of the core changed to add this
support, but it needs to be justified.  I don't want to
support multiple different ways of doing the same thing.

Maybe we should backport the changes in 4.1 to our tree
before we start hacking into it?  We could leave out the adp
changes etc. as I believe someone said they were broken.  We
should make a point release before doing this.  What do you
think?



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 05:46

Message:
Logged In: YES 
user_id=87254


Sorry, I don't have a patch to show how UDP/Unix sockets
could be integrated into nssock.  The 4.1 code base would
seem to be a better starting point as it creates a new
thread for each loaded comm driver.

Here's how it could work:

Unix domain socket support is easy.  Use the same syntax as
for the binder for the address param of the driver module in
the config file.  Call the correct socket listen routine
depending on type within NsStartDrivers().  The nssock
module should require no changes to work with this socket type.

UDP sockets are harder.  The 4.1 code base has changed
NsStartDrivers() to spin off a thread for each loaded comm
module.  As the requirements for stream and packet sockets
are different, a different routine would be passed to
Ns_ThreadCreate() for each type.

I would rename DriverThread() StreamThread() and add a new
routine, PacketThread().  Much of the contents of
DriverThread() can be junked for handling packet based
protocols, but not all should.  The new QueueWait callbacks
and pre-queue filters are just as applicable here as for
stream based protocols.

The advantage of this is that there is a clear distinction
between network transport, protocol, and application. 
Network transports such as SSL/TLS, SSH, etc. can be swapped
in below different protocol implementations.  Many
applications can be built on a protocol implementation if it
provides the standard hooks (registered proc, filters,
QueueWait callbacks, pre-queue filteres...).

Is this enough to give you an idea?  It seems feasable to
me, what do you think?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 21:31

Message:
Logged In: YES 
user_id=184124

Do we have any conclusions about this? Soon i will have to
develop some application with support for SIP over UDP, i
would like to do it on Naviserver and not on my hacked AS.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 09:44

Message:
Logged In: YES 
user_id=184124

And again, i have nothing against it, i said it before, what
i am offering is a littl ebit different, for network based
protocols driver stuff is kay, i just want to be able to
submit conn threads from within the server, not through the
main driver.
If i have already request data, through shared memory, file
or other way, i want to submit wroker thread directly, not
by connecting to myself and submitting data using HTTP or
other protocol i will have to implement just to queue te
request. This is for other applications, not HTTP/Web based,
more like messaging/authentication servers and alike.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 07:41

Message:
Logged In: YES 
user_id=184124

Can you provide what you have to modify in driver/nssock, i
think that will not be easy and will require too many
modifications.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-04 03:08

Message:
Logged In: YES 
user_id=87254


What do you think of my suggestion that the nssock module
and nsd/driver.c could be easily modified to handle udp and
unix domain sockets when specified in the config file?

A change like that combined with the work you've already
done to the binder would mean no development is required to
handle all socket types, and would be transparent to the
upper layers.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-26 10:10

Message:
Logged In: YES 
user_id=184124

Stephan,
I am suggesting a compromise, let's put ParseProc and
DriverSockQueu patches, they can coexists easily but will
offer two way to add new drivers. Once we have new drivers
working we will see what can be modified/adjusted, for now
we do not have anything except my several drivers i use in
the production and i use my own loop for smtp driver to keep
processing and state machine in one place.
Again, i have nothing against your method, i used to do
similar in my previous versions but now i need to be
completely free to implement aditional driver and i think we
should have this ability.

Zoran, what do you think?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 15:57

Message:
Logged In: YES 
user_id=184124

Try to implement HTTP over UDP as i provided in the patch,
you will need to modify core driver again, this is one example.

Next, if i want to implement small driver that do not
interact with web part of aolserver, no filters, it just
need to do one small thing, it should be fast but will
handle very many requests. I do not need filters, callbacks,
i need my own main loop. 
I do not understand why you are insisting that all drivers
should go through http driver's main loop. If i implement
dns server and web interface to it, why they should compete
in the same driver's main loop and go through registered
filters if they share only backend database structure?

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 15:22

Message:
Logged In: YES 
user_id=87254


You don't have to use registered filters or procs.  It's an
option for protocol implementations where the initial accept
and parsing take place in the driver thread, and processing
takes place in a conn thread.

Once you submit to the queue however, either a filter, a
registered proc, or a registered proxy proc must handle the
request.  It makes no difference whether you submit
explicitly with Ns_DriverSockQueue or implicitly after your
Ns_ParseProc completes.

How much of the request structure you fill in is up to you,
it's all optional.  If you choose not to use it there is no
overhead.  It doesn't have to be a text based protocol for
this to make sense.  You might fill in the request structure
if you wanted to enable people to handle different request
types via C or Tcl.  Or, you could store binary data in
Ns_Cls storage and access it from registered procs etc.

In the case where you want to handle connection
accept/parse/reponse processing in the same thread, nothing
needs to be added.  You can do that today with
Ns_SockCallback (or by placing all your code in an
Ns_ParseProc and returning the response from there)

Can you be more specific about the overhead of using an
Ns_ParseProc?  Exactly what memory is allocated, what code
is run that should not..?

What specifically can you not do with the Ns_ParseProc
interface that you need the Ns_Driver* routines for?  I'm
looking for concrete examples.

As far as I can tell, with Ns_ParseProc you write less code,
get more options, and it takes advantage of infrastructure
to give you more speed.  I must be missing something, but
you'll have to explain it to me in more detail because I
don't get it.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 13:43

Message:
Logged In: YES 
user_id=184124

Okay, the biggest problem with your way i see that i am
enforced to go through http driver, even with my parse proc,
i need to know how it works to implement my driver so
correct hooks/filters will be called. If i want skip some
parts i am not able to do it, http driver works only one way. 
Another issue is if something will not fit into current
driver, additional hook needs to be introduced and core
driver needs to be modified again.

If my driver fits http-like/text mode paradigm, reusing http
driver is the easiest way, but if i need something specific,
all extra step to mimic the connection as http request, so 
all other parts of the http driver will not fail is just
unnecesssary extra efforts.

And i you mentioned, using Ns_DriverSockQueue is low level
function and requires from the developer full attention for
bulding the driver, but if this is what i want, why not. I
can spawn my own thread if some callbacks are long running
Tcl scripts and queue connection from my own loop. If i omly
want to received packets, decode them and submit connection,
i do not need all http driver infrusturcture, why to enforce
using it?

So we can have at least 3 ways for supporting multiprotocols:
- standard callbacks, completely different threads, no
connection pooling (Exists now)

- Ns_DriverSockQueue, new thread or callback thread, reusing
connection pooling. All filters/traces are reused as well

- driver Parse proc in the http driver to reuse drivers
thread and connection pooling with all filters/traces

All 3 methiods can co-exists and do not interfere with each
other, that's my point, and all they add just couple extra
functions. I would implement all methods, i have 3 drivers i
need to run: smtp, dns, snmp, can add imap in the future

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 13:31

Message:
Logged In: YES 
user_id=87254


"Each way has its own drawbacks..."

That's what I'm asking.  What are the drawbacks for each
approach we've come up with so far?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 08:23

Message:
Logged In: YES 
user_id=184124

I am not arguing with you about your ParseProc driver
extension, it is very usefull and makes driver more flexible.

I just want to add couple of new API functions that will allow
developer to queue connections from any place, that's it.
How drivers will be written and how developer will decide to
handle it is up to each particular developer. But nobody
will be locked up in only one way of doing it.

Each way has its own drawbacks so we have multiple choices
and developer will decide for itself how it should be.

Both our ways do not mutually exclusive.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 01:28

Message:
Logged In: YES 
user_id=87254


Ah I see you're right, an extra thread is not created for
each incoming request.  'udpThread' is perhaps not the best
name for the socket callback though... :-)

I still don't see the advantage of using Ns_SockCalback.  A
single thread is created by the server to handle all
registered callbacks, including those from ns_sockcallback.
 At runtime, while this single thread is running tcl code to
handle one callback, your dirver callback cannot run, no new
connections will be accepted, etc.  Why is it better to use
the Ns_SockCallback thread rather than a driver thread?

Maybe I'm reading this wrong, but how do you handle the case
where the request arrives as more than one packet? 
Ns_DriverSockRead() is called from the 'udpThread' socket
callback.  The only return value checked is NS_OK, but
couldn't this also be SOCK_MORE?

How would you handle things like keepalive?  I think you'd
have to reimplement that in the Ns_SockCallback thread.

Re the proxy stuff, Ns_RegisterRequest() and
Ns_RegisterProxyRequest() seem very simillar.  With
Ns_RegisterRequest(), filters are run and you get the choice
of using C or Tcl.  Ns_RegisterProxyRequest() offers no
advantage that I can see -- you still need a complete
Request structure, even if you just ignore it.


The comm driver initialization should probably be changed to
automatically handle unix domain sockets.  You add /foo
rather that 127.0.0.1 in the config file and it knows to
create the correct type of socket.  nssock and nsopenssl
wouldn't have to be modified at all.

UDP is different.  There are two types of protocols: single
packet, such as DNS or RADIUS; multi packet, as used in some
streaming media and p2p protocols.

I think for the single packet case we again might want to
modify the driver code to automatically handle it.  No
read-ahead is necessary, there's only one packet per
request, so it's placed in the request buffer and passed on
to the next stage, which is parsing.  Here, a return code of
SOCK_MORE would be illegal.

Every multi-packet UDP protocol will require custom
framing/sequencing and the developer will have to create a
new socket driver.

Taking RADIUS as an example, which is a single packet UDP
protocol, you'd create a very simple Ns_ParseProc who's only
job is to check that the number of bytes specified in the
header actually arrived, and return SOCK_READY.  A default
request structure is created for you so the only other thing
you have to do is Ns_RegisterRequest() for the '/' URL. 
Within your request proc, call Ns_ConnContent() and parse
the buffer.

Now, you do have a number of other options to make this more
flexible.  You could parse the request in your Ns_ParseProc
and then fill out the request structure.  e.g. the different
RADIUS message types could be expressed as HTTP verbs.  This
buys you flexibility.  Now you can Ns_RegisterRequest() a
different routine for each RADIUS message type, and someone
reusing your code can override your default implementation.
 You can also handle some message types in C, and others in Tcl.

You might decide to put some useful information about the
RADIUS request in the URL line.  Now you get logging for free.

You might decide to parse the key/value pairs from the
RADIUS request into query vars or headers.  Now you don't
have to write a bunch of RADIUS specific Tcl commands to
access the data, should you want to handle that from Tcl. 
ns_queryget or ns_conn headers  will work.

How far you want to go is up to you, the writer of the
protocol module.  But at the low end all you need is an
Ns_ParseProc and usually a single registered request proc,
which is a lot less code to write than the Ns_Driver* stuff,
I think.

I'm sure it's not perfect!  So what's wrong with it? Why
would it suck to write a RADIUS protocol module as I've just
described, to take one example?





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:51

Message:
Logged In: YES 
user_id=184124

I think we need to collect all solutions and then see where
are going, i am holding to port all my old/running modules
because i do not know how naviserver will handle foreign
protocols.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:48

Message:
Logged In: YES 
user_id=184124

yes, using proxy can solve immediate requirements without
hacking NsConnProc by adding hooks to call driver specific C
functions.

If i need my smtp server and main loop is in C, i need
somehow call it
in the connection thread. Using registered proxy function i
can do it now, i do not need filter/traces. This is for
completely new modules.

I can implement main loop in the module as Tcl command and
then call it in the connection filter, it is possible, it
will just require many different parts to be in place and
still filter should be registered as Tcl proc which will
call another driver main loop.

sometimes low level stuff makes things easier and simpler:-)))

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:45

Message:
Logged In: YES 
user_id=184124

In all my installations i need sha module but it wasn't in
the distribution, so i need to download it or repackage
aolserver to include nssha1. That is my point, it is just
one simple function use more often than something like
ns_jpegsize or so.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-24 22:42

Message:
Logged In: YES 
user_id=87254


Well I don't know about the wisdom of adding sha1 to the
core at this stage  :-)

I see your point though, encryption and hashing functions
are almost universally required for systems/server work. 
Maybe we need to consider adding a new encryption module to
the core distribution.  Like nsdb it would export a C API
via libnscypher.so (or whatever) as well as the Tcl module
nscypher.so.

Times have changed and things like the openssl libraries are
common on all platforms so it's not the big deal it used to
be to add such dependiencies.

Such a module should make implementing SSL in nsopenssl easier.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:41

Message:
Logged In: YES 
user_id=184124

No, i do not spawn new thread, i use callback feature of the
server, when socket is ready, server calls provided callback
ad that callback just submits the socket to connection pool.
if pool is full, Ns_DriverSockQueue will return NS_TIMEOUT,
you can retry.
Yes, it is low-level, for high level, HTTP driver provides a
lot of functionality, it could be extended like you did, but
still it is HTTP driver hacked. If i want completely new
driver, like RADIUS server, http driver will not help me, i
need low level stuff, and it is there already.

To reuse resource limiting, i added Ns_DriverSockQueue
function, so
new conections can be queue instead of creating new threads.

I do not have anything against your patch for extending
current driver, it is very usefull. i am offering new API
for low level drivers.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-24 22:36

Message:
Logged In: YES 
user_id=87254


The proxy stuff is too late in the cycle to do much.  The
request has to be fully parsed by then (for read-ahead).

If a proxy function is available, then all filters, the auth
phase, registered procs, cleanup procs etc. are bypassed.  A
lot of that stuff can be very useful for non-HTTP protocols.

I don't think handling stuff via a proxy function buys you much.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-24 22:32

Message:
Logged In: YES 
user_id=87254


I don't understand what you're trying to acheive here (well,
apart from multi-protocol... :-).  The newly exposed
Ns_Driver* entry points are quite low level, and so the
implementor of the new protocol is left to do a lot of the
heavy lifting.

For example, the way I read it you have to create your own
listen socket and register a callback.  Every time a new
reaquest comes in a thread is spawned to handle it.  From
that thread you then submit the parsed request to one of the
conn threads.

Excessive thread creation and message passing between
threads is not going to perform well.  And it seems you have
to write more code than e.g. the example POP3 driver I
posted some time ago.

You're also not taking advantage of the other facilities
that the server offers.  What happens if 1000 connections
arrive, do you spawn 1000 threads?  You could of course code
up some limit checks, but this already exists.  What if a
client sends you a continuous stream of data, 2GB...  etc.

By using the driver hooks to provide the new protocol
parser, you deny yourself the opportunity to use something
like the nsopenssl module.  This should work just fine for
protocols like SMTP, IMAP, POP3 and probably others.


Anyway, I think one of the most carefully coded aspects of
the server is it's attention to resource usage.  That goes
for IO, context switching, memory, etc.  It's espescialy
nice that most of the time you're not even aware that all
this work is being done for you.  I'd like new protocol
drivers to be able to transparently take advantage of that.

Could you take a look at my old POP3 demo driver?  It's the
attachment nspopd-0.3.tar.bz2 over here:

http://sourceforge.net/tracker/index.php?func=detail&aid=973010&group_id=3152&atid=353152

It's not obvious that anything interesting is going on, so
it's not much to look at.  But actually, conn socket
read-ahead is happening eficiently in one thread with async
IO, the conn threads are treated as a precious resource
(heavy-weight Tcl interps) and are allocated at the last
minute, there's an easy API in C and Tcl to implement the
actual reading of data from the INBOX (could be from the
file system, db etc.).

You've got a lot of experience writing servers, what do you
think is wrong with this model?  What can it not do, or what
could it do better?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 14:29

Message:
Logged In: YES 
user_id=184124

Another thing, once we can submit connections from any
place, no need to build any drivers, even in C, i can
register new proxy proc and set protocol field in my
request, so when submitted, connection will run registered
proxy proc.
for example: 
 in my smtp driver/module, i create driver, register proxy
for smtp: protocol, register callback for the socket. Once
connection accepted, in my module i submit that connection
to the queue with request-protocl set to smtp:.
queue.c will call my proxy handler, which is C function.

No need to add anything else. This way even standard
aolserver can be extended without touching precious http
driver thread. Sorry for sarcasm.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

[ naviserver-Feature Requests-1151137 ] New driver API and Udp module

[SourceForge.net <no...@so...>]

Feature Requests item #1151137, was opened at 2005-02-24 09:01
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

Category: C-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Vlad Seryakov (seryakov)
Assigned to: Vlad Seryakov (seryakov)
Summary: New driver API and Udp module

Initial Comment:
Hi guys,

Attached is minor driver extensions which do not change
existing drivers  but add new functionality. There are
some cosmetic changes, like moving some fields in the
Ns_Sock/Ns_Driver structres so they can be accessed
publically and made some private functions public but
functionality preserved as before.

I included udp driver as an example of new API, and
also added ns_sha1 command in the tclmisc.c, it is just
one command and it is uses practically everywhere.

-------------------------------------------------------------------------
To test udp driver i use new ns_udp command:

ossweb:nscp 8> ns_udp send 127.0.0.1 5060 "GET /
HTTP/1.0\n\n"

HTTP/1.0 200 OK
MIME-Version: 1.0
Date: Thu, 24 Feb 2005 05:39:50 GMT
Server: NaviServer/4.0.10
Content-Type: text/html; charset=iso-8859-1
Content-Length: 661
Connection: close

<HEAD><TITLE>Seryakov's Family Intranet</TITLE></HEAD>
.....


----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 05:46

Message:
Logged In: YES 
user_id=87254


Sorry, I don't have a patch to show how UDP/Unix sockets
could be integrated into nssock.  The 4.1 code base would
seem to be a better starting point as it creates a new
thread for each loaded comm driver.

Here's how it could work:

Unix domain socket support is easy.  Use the same syntax as
for the binder for the address param of the driver module in
the config file.  Call the correct socket listen routine
depending on type within NsStartDrivers().  The nssock
module should require no changes to work with this socket type.

UDP sockets are harder.  The 4.1 code base has changed
NsStartDrivers() to spin off a thread for each loaded comm
module.  As the requirements for stream and packet sockets
are different, a different routine would be passed to
Ns_ThreadCreate() for each type.

I would rename DriverThread() StreamThread() and add a new
routine, PacketThread().  Much of the contents of
DriverThread() can be junked for handling packet based
protocols, but not all should.  The new QueueWait callbacks
and pre-queue filters are just as applicable here as for
stream based protocols.

The advantage of this is that there is a clear distinction
between network transport, protocol, and application. 
Network transports such as SSL/TLS, SSH, etc. can be swapped
in below different protocol implementations.  Many
applications can be built on a protocol implementation if it
provides the standard hooks (registered proc, filters,
QueueWait callbacks, pre-queue filteres...).

Is this enough to give you an idea?  It seems feasable to
me, what do you think?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 21:31

Message:
Logged In: YES 
user_id=184124

Do we have any conclusions about this? Soon i will have to
develop some application with support for SIP over UDP, i
would like to do it on Naviserver and not on my hacked AS.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 09:44

Message:
Logged In: YES 
user_id=184124

And again, i have nothing against it, i said it before, what
i am offering is a littl ebit different, for network based
protocols driver stuff is kay, i just want to be able to
submit conn threads from within the server, not through the
main driver.
If i have already request data, through shared memory, file
or other way, i want to submit wroker thread directly, not
by connecting to myself and submitting data using HTTP or
other protocol i will have to implement just to queue te
request. This is for other applications, not HTTP/Web based,
more like messaging/authentication servers and alike.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-04 07:41

Message:
Logged In: YES 
user_id=184124

Can you provide what you have to modify in driver/nssock, i
think that will not be easy and will require too many
modifications.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-04 03:08

Message:
Logged In: YES 
user_id=87254


What do you think of my suggestion that the nssock module
and nsd/driver.c could be easily modified to handle udp and
unix domain sockets when specified in the config file?

A change like that combined with the work you've already
done to the binder would mean no development is required to
handle all socket types, and would be transparent to the
upper layers.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-26 10:10

Message:
Logged In: YES 
user_id=184124

Stephan,
I am suggesting a compromise, let's put ParseProc and
DriverSockQueu patches, they can coexists easily but will
offer two way to add new drivers. Once we have new drivers
working we will see what can be modified/adjusted, for now
we do not have anything except my several drivers i use in
the production and i use my own loop for smtp driver to keep
processing and state machine in one place.
Again, i have nothing against your method, i used to do
similar in my previous versions but now i need to be
completely free to implement aditional driver and i think we
should have this ability.

Zoran, what do you think?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 15:57

Message:
Logged In: YES 
user_id=184124

Try to implement HTTP over UDP as i provided in the patch,
you will need to modify core driver again, this is one example.

Next, if i want to implement small driver that do not
interact with web part of aolserver, no filters, it just
need to do one small thing, it should be fast but will
handle very many requests. I do not need filters, callbacks,
i need my own main loop. 
I do not understand why you are insisting that all drivers
should go through http driver's main loop. If i implement
dns server and web interface to it, why they should compete
in the same driver's main loop and go through registered
filters if they share only backend database structure?

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 15:22

Message:
Logged In: YES 
user_id=87254


You don't have to use registered filters or procs.  It's an
option for protocol implementations where the initial accept
and parsing take place in the driver thread, and processing
takes place in a conn thread.

Once you submit to the queue however, either a filter, a
registered proc, or a registered proxy proc must handle the
request.  It makes no difference whether you submit
explicitly with Ns_DriverSockQueue or implicitly after your
Ns_ParseProc completes.

How much of the request structure you fill in is up to you,
it's all optional.  If you choose not to use it there is no
overhead.  It doesn't have to be a text based protocol for
this to make sense.  You might fill in the request structure
if you wanted to enable people to handle different request
types via C or Tcl.  Or, you could store binary data in
Ns_Cls storage and access it from registered procs etc.

In the case where you want to handle connection
accept/parse/reponse processing in the same thread, nothing
needs to be added.  You can do that today with
Ns_SockCallback (or by placing all your code in an
Ns_ParseProc and returning the response from there)

Can you be more specific about the overhead of using an
Ns_ParseProc?  Exactly what memory is allocated, what code
is run that should not..?

What specifically can you not do with the Ns_ParseProc
interface that you need the Ns_Driver* routines for?  I'm
looking for concrete examples.

As far as I can tell, with Ns_ParseProc you write less code,
get more options, and it takes advantage of infrastructure
to give you more speed.  I must be missing something, but
you'll have to explain it to me in more detail because I
don't get it.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 13:43

Message:
Logged In: YES 
user_id=184124

Okay, the biggest problem with your way i see that i am
enforced to go through http driver, even with my parse proc,
i need to know how it works to implement my driver so
correct hooks/filters will be called. If i want skip some
parts i am not able to do it, http driver works only one way. 
Another issue is if something will not fit into current
driver, additional hook needs to be introduced and core
driver needs to be modified again.

If my driver fits http-like/text mode paradigm, reusing http
driver is the easiest way, but if i need something specific,
all extra step to mimic the connection as http request, so 
all other parts of the http driver will not fail is just
unnecesssary extra efforts.

And i you mentioned, using Ns_DriverSockQueue is low level
function and requires from the developer full attention for
bulding the driver, but if this is what i want, why not. I
can spawn my own thread if some callbacks are long running
Tcl scripts and queue connection from my own loop. If i omly
want to received packets, decode them and submit connection,
i do not need all http driver infrusturcture, why to enforce
using it?

So we can have at least 3 ways for supporting multiprotocols:
- standard callbacks, completely different threads, no
connection pooling (Exists now)

- Ns_DriverSockQueue, new thread or callback thread, reusing
connection pooling. All filters/traces are reused as well

- driver Parse proc in the http driver to reuse drivers
thread and connection pooling with all filters/traces

All 3 methiods can co-exists and do not interfere with each
other, that's my point, and all they add just couple extra
functions. I would implement all methods, i have 3 drivers i
need to run: smtp, dns, snmp, can add imap in the future

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 13:31

Message:
Logged In: YES 
user_id=87254


"Each way has its own drawbacks..."

That's what I'm asking.  What are the drawbacks for each
approach we've come up with so far?

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-25 08:23

Message:
Logged In: YES 
user_id=184124

I am not arguing with you about your ParseProc driver
extension, it is very usefull and makes driver more flexible.

I just want to add couple of new API functions that will allow
developer to queue connections from any place, that's it.
How drivers will be written and how developer will decide to
handle it is up to each particular developer. But nobody
will be locked up in only one way of doing it.

Each way has its own drawbacks so we have multiple choices
and developer will decide for itself how it should be.

Both our ways do not mutually exclusive.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-25 01:28

Message:
Logged In: YES 
user_id=87254


Ah I see you're right, an extra thread is not created for
each incoming request.  'udpThread' is perhaps not the best
name for the socket callback though... :-)

I still don't see the advantage of using Ns_SockCalback.  A
single thread is created by the server to handle all
registered callbacks, including those from ns_sockcallback.
 At runtime, while this single thread is running tcl code to
handle one callback, your dirver callback cannot run, no new
connections will be accepted, etc.  Why is it better to use
the Ns_SockCallback thread rather than a driver thread?

Maybe I'm reading this wrong, but how do you handle the case
where the request arrives as more than one packet? 
Ns_DriverSockRead() is called from the 'udpThread' socket
callback.  The only return value checked is NS_OK, but
couldn't this also be SOCK_MORE?

How would you handle things like keepalive?  I think you'd
have to reimplement that in the Ns_SockCallback thread.

Re the proxy stuff, Ns_RegisterRequest() and
Ns_RegisterProxyRequest() seem very simillar.  With
Ns_RegisterRequest(), filters are run and you get the choice
of using C or Tcl.  Ns_RegisterProxyRequest() offers no
advantage that I can see -- you still need a complete
Request structure, even if you just ignore it.


The comm driver initialization should probably be changed to
automatically handle unix domain sockets.  You add /foo
rather that 127.0.0.1 in the config file and it knows to
create the correct type of socket.  nssock and nsopenssl
wouldn't have to be modified at all.

UDP is different.  There are two types of protocols: single
packet, such as DNS or RADIUS; multi packet, as used in some
streaming media and p2p protocols.

I think for the single packet case we again might want to
modify the driver code to automatically handle it.  No
read-ahead is necessary, there's only one packet per
request, so it's placed in the request buffer and passed on
to the next stage, which is parsing.  Here, a return code of
SOCK_MORE would be illegal.

Every multi-packet UDP protocol will require custom
framing/sequencing and the developer will have to create a
new socket driver.

Taking RADIUS as an example, which is a single packet UDP
protocol, you'd create a very simple Ns_ParseProc who's only
job is to check that the number of bytes specified in the
header actually arrived, and return SOCK_READY.  A default
request structure is created for you so the only other thing
you have to do is Ns_RegisterRequest() for the '/' URL. 
Within your request proc, call Ns_ConnContent() and parse
the buffer.

Now, you do have a number of other options to make this more
flexible.  You could parse the request in your Ns_ParseProc
and then fill out the request structure.  e.g. the different
RADIUS message types could be expressed as HTTP verbs.  This
buys you flexibility.  Now you can Ns_RegisterRequest() a
different routine for each RADIUS message type, and someone
reusing your code can override your default implementation.
 You can also handle some message types in C, and others in Tcl.

You might decide to put some useful information about the
RADIUS request in the URL line.  Now you get logging for free.

You might decide to parse the key/value pairs from the
RADIUS request into query vars or headers.  Now you don't
have to write a bunch of RADIUS specific Tcl commands to
access the data, should you want to handle that from Tcl. 
ns_queryget or ns_conn headers  will work.

How far you want to go is up to you, the writer of the
protocol module.  But at the low end all you need is an
Ns_ParseProc and usually a single registered request proc,
which is a lot less code to write than the Ns_Driver* stuff,
I think.

I'm sure it's not perfect!  So what's wrong with it? Why
would it suck to write a RADIUS protocol module as I've just
described, to take one example?





----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:51

Message:
Logged In: YES 
user_id=184124

I think we need to collect all solutions and then see where
are going, i am holding to port all my old/running modules
because i do not know how naviserver will handle foreign
protocols.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:48

Message:
Logged In: YES 
user_id=184124

yes, using proxy can solve immediate requirements without
hacking NsConnProc by adding hooks to call driver specific C
functions.

If i need my smtp server and main loop is in C, i need
somehow call it
in the connection thread. Using registered proxy function i
can do it now, i do not need filter/traces. This is for
completely new modules.

I can implement main loop in the module as Tcl command and
then call it in the connection filter, it is possible, it
will just require many different parts to be in place and
still filter should be registered as Tcl proc which will
call another driver main loop.

sometimes low level stuff makes things easier and simpler:-)))

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:45

Message:
Logged In: YES 
user_id=184124

In all my installations i need sha module but it wasn't in
the distribution, so i need to download it or repackage
aolserver to include nssha1. That is my point, it is just
one simple function use more often than something like
ns_jpegsize or so.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-24 22:42

Message:
Logged In: YES 
user_id=87254


Well I don't know about the wisdom of adding sha1 to the
core at this stage  :-)

I see your point though, encryption and hashing functions
are almost universally required for systems/server work. 
Maybe we need to consider adding a new encryption module to
the core distribution.  Like nsdb it would export a C API
via libnscypher.so (or whatever) as well as the Tcl module
nscypher.so.

Times have changed and things like the openssl libraries are
common on all platforms so it's not the big deal it used to
be to add such dependiencies.

Such a module should make implementing SSL in nsopenssl easier.



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 22:41

Message:
Logged In: YES 
user_id=184124

No, i do not spawn new thread, i use callback feature of the
server, when socket is ready, server calls provided callback
ad that callback just submits the socket to connection pool.
if pool is full, Ns_DriverSockQueue will return NS_TIMEOUT,
you can retry.
Yes, it is low-level, for high level, HTTP driver provides a
lot of functionality, it could be extended like you did, but
still it is HTTP driver hacked. If i want completely new
driver, like RADIUS server, http driver will not help me, i
need low level stuff, and it is there already.

To reuse resource limiting, i added Ns_DriverSockQueue
function, so
new conections can be queue instead of creating new threads.

I do not have anything against your patch for extending
current driver, it is very usefull. i am offering new API
for low level drivers.

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-24 22:36

Message:
Logged In: YES 
user_id=87254


The proxy stuff is too late in the cycle to do much.  The
request has to be fully parsed by then (for read-ahead).

If a proxy function is available, then all filters, the auth
phase, registered procs, cleanup procs etc. are bypassed.  A
lot of that stuff can be very useful for non-HTTP protocols.

I don't think handling stuff via a proxy function buys you much.



----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-02-24 22:32

Message:
Logged In: YES 
user_id=87254


I don't understand what you're trying to acheive here (well,
apart from multi-protocol... :-).  The newly exposed
Ns_Driver* entry points are quite low level, and so the
implementor of the new protocol is left to do a lot of the
heavy lifting.

For example, the way I read it you have to create your own
listen socket and register a callback.  Every time a new
reaquest comes in a thread is spawned to handle it.  From
that thread you then submit the parsed request to one of the
conn threads.

Excessive thread creation and message passing between
threads is not going to perform well.  And it seems you have
to write more code than e.g. the example POP3 driver I
posted some time ago.

You're also not taking advantage of the other facilities
that the server offers.  What happens if 1000 connections
arrive, do you spawn 1000 threads?  You could of course code
up some limit checks, but this already exists.  What if a
client sends you a continuous stream of data, 2GB...  etc.

By using the driver hooks to provide the new protocol
parser, you deny yourself the opportunity to use something
like the nsopenssl module.  This should work just fine for
protocols like SMTP, IMAP, POP3 and probably others.


Anyway, I think one of the most carefully coded aspects of
the server is it's attention to resource usage.  That goes
for IO, context switching, memory, etc.  It's espescialy
nice that most of the time you're not even aware that all
this work is being done for you.  I'd like new protocol
drivers to be able to transparently take advantage of that.

Could you take a look at my old POP3 demo driver?  It's the
attachment nspopd-0.3.tar.bz2 over here:

http://sourceforge.net/tracker/index.php?func=detail&aid=973010&group_id=3152&atid=353152

It's not obvious that anything interesting is going on, so
it's not much to look at.  But actually, conn socket
read-ahead is happening eficiently in one thread with async
IO, the conn threads are treated as a precious resource
(heavy-weight Tcl interps) and are allocated at the last
minute, there's an easy API in C and Tcl to implement the
actual reading of data from the INBOX (could be from the
file system, db etc.).

You've got a lot of experience writing servers, what do you
think is wrong with this model?  What can it not do, or what
could it do better?



----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-02-24 14:29

Message:
Logged In: YES 
user_id=184124

Another thing, once we can submit connections from any
place, no need to build any drivers, even in C, i can
register new proxy proc and set protocol field in my
request, so when submitted, connection will run registered
proxy proc.
for example: 
 in my smtp driver/module, i create driver, register proxy
for smtp: protocol, register callback for the socket. Once
connection accepted, in my module i submit that connection
to the queue with request-protocl set to smtp:.
queue.c will call my proxy handler, which is C function.

No need to add anything else. This way even standard
aolserver can be extended without touching precious http
driver thread. Sorry for sarcasm.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1151137&group_id=130646

[ naviserver-Feature Requests-1159259 ] Add nscache module functionality

[SourceForge.net <no...@so...>]

Feature Requests item #1159259, was opened at 2005-03-08 19:30
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159259&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Stephen Deasey (sdeasey)
Summary: Add nscache module functionality

Initial Comment:
This is really what needs to be done sine the server itself
does support the caches and what lacks is the Tcl access
to them. The nscache module (or its equivalent) deserves
to be part of the server distribution. 

----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 13:15

Message:
Logged In: YES 
user_id=95086

Take whatever time you need. I do not have any problems 
with that. 
I also think that namespaced commands are better, so 
instead of 
 
  ns_cache blabla 
 
the better way would be 
 
  ns::cache::blabla 
 
but I think it is to early to make this kind of switch. We can 
eventually make this sometime, leaving the ns_xxxx for 
compatibility, but this is another story. 

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 13:03

Message:
Logged In: YES 
user_id=87254


The existing nscache module is a good example of the
ensemble method of command parsing sucking, IMHO.  The
master command has to parse the first word to figure out
which subcommand to invoke, and then dispatches to the
implementation of that subcommand as they have nothing in
common.  It requires extra code and slows performance.

There are locking problems with the current API. 
ns_cache_stats, for example, can access any cache in the
system at any time but there is no locking of the Cache
structures.

The built in commands work on a global level, the commands
in the nscache module operate only on caches created with
those commands.  It would be a shame to junk the ability of
the current commands to work with all caches.

The nscache module works fine as is, and  as I've already
started work on integrating the tcl cache commands I'd
prefer if the old ones weren't added in the mean time.  Can
you give me some time to post a patch to the original cache RFE?



----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 09:56

Message:
Logged In: YES 
user_id=95086

I would also somehow unify the Tcl api interface. 
If you look at the server part it does: 
 
ns_cache_XXXXX 
 
whereas the cache module has different api 
 
ns_cache YYYY 
 
I prefer the latter (ensemble-based) api. If we move 
the cache module in the core, we should make  
  ns_cache YYYYY 
wrappers to do the ns_cache_XXXXX commands. 
What do you think? 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 05:33

Message:
Logged In: YES 
user_id=184124

For a start we can just add tclcache.c into the core and
leter improve/rewrite caching if needed. Currenlty i am
pretty happy with this module functionality as is.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-08 20:48

Message:
Logged In: YES 
user_id=184124

Yes, i agree with it

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159259&group_id=130646

[ naviserver-Feature Requests-1159259 ] Add nscache module functionality

[SourceForge.net <no...@so...>]

Feature Requests item #1159259, was opened at 2005-03-08 11:30
Message generated for change (Comment added) made by sdeasey
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159259&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Stephen Deasey (sdeasey)
Summary: Add nscache module functionality

Initial Comment:
This is really what needs to be done sine the server itself
does support the caches and what lacks is the Tcl access
to them. The nscache module (or its equivalent) deserves
to be part of the server distribution. 

----------------------------------------------------------------------

>Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 05:03

Message:
Logged In: YES 
user_id=87254


The existing nscache module is a good example of the
ensemble method of command parsing sucking, IMHO.  The
master command has to parse the first word to figure out
which subcommand to invoke, and then dispatches to the
implementation of that subcommand as they have nothing in
common.  It requires extra code and slows performance.

There are locking problems with the current API. 
ns_cache_stats, for example, can access any cache in the
system at any time but there is no locking of the Cache
structures.

The built in commands work on a global level, the commands
in the nscache module operate only on caches created with
those commands.  It would be a shame to junk the ability of
the current commands to work with all caches.

The nscache module works fine as is, and  as I've already
started work on integrating the tcl cache commands I'd
prefer if the old ones weren't added in the mean time.  Can
you give me some time to post a patch to the original cache RFE?



----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 01:56

Message:
Logged In: YES 
user_id=95086

I would also somehow unify the Tcl api interface. 
If you look at the server part it does: 
 
ns_cache_XXXXX 
 
whereas the cache module has different api 
 
ns_cache YYYY 
 
I prefer the latter (ensemble-based) api. If we move 
the cache module in the core, we should make  
  ns_cache YYYYY 
wrappers to do the ns_cache_XXXXX commands. 
What do you think? 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-17 21:33

Message:
Logged In: YES 
user_id=184124

For a start we can just add tclcache.c into the core and
leter improve/rewrite caching if needed. Currenlty i am
pretty happy with this module functionality as is.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-08 12:48

Message:
Logged In: YES 
user_id=184124

Yes, i agree with it

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159259&group_id=130646

[ naviserver-Feature Requests-1159259 ] Add nscache module functionality

[SourceForge.net <no...@so...>]

Feature Requests item #1159259, was opened at 2005-03-08 19:30
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159259&group_id=130646

Category: Tcl-API
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Stephen Deasey (sdeasey)
Summary: Add nscache module functionality

Initial Comment:
This is really what needs to be done sine the server itself
does support the caches and what lacks is the Tcl access
to them. The nscache module (or its equivalent) deserves
to be part of the server distribution. 

----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 09:56

Message:
Logged In: YES 
user_id=95086

I would also somehow unify the Tcl api interface. 
If you look at the server part it does: 
 
ns_cache_XXXXX 
 
whereas the cache module has different api 
 
ns_cache YYYY 
 
I prefer the latter (ensemble-based) api. If we move 
the cache module in the core, we should make  
  ns_cache YYYYY 
wrappers to do the ns_cache_XXXXX commands. 
What do you think? 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 05:33

Message:
Logged In: YES 
user_id=184124

For a start we can just add tclcache.c into the core and
leter improve/rewrite caching if needed. Currenlty i am
pretty happy with this module functionality as is.

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-08 20:48

Message:
Logged In: YES 
user_id=184124

Yes, i agree with it

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1159259&group_id=130646

[ naviserver-Feature Requests-1165562 ] Scrap aolpress handling

[SourceForge.net <no...@so...>]

Feature Requests item #1165562, was opened at 2005-03-17 21:50
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1165562&group_id=130646

Category: C-API
Group: None
>Status: Closed
>Resolution: Accepted
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Zoran Vasiljevic (vasiljevic)
Summary: Scrap aolpress handling

Initial Comment:
In Ns_ConnSetRequiredHeaders() there is a provision to add
NaviServer/2.0 to headers. I doubt that anybody will ever 
need this any more. I'd scrap this entirely.

----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 09:41

Message:
Logged In: YES 
user_id=95086

Removed. 

----------------------------------------------------------------------

Comment By: Vlad Seryakov (seryakov)
Date: 2005-03-18 05:28

Message:
Logged In: YES 
user_id=184124

There are also places in server.c , nsd.h and nsdconf.h with
aolpress. Let's remove them

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 01:21

Message:
Logged In: YES 
user_id=87254


There's more aolpress junk in tcl/fastpath.tcl which can be
removed at the same time.

I think this closed-source client is officially dead...


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1165562&group_id=130646

[ naviserver-Feature Requests-1156899 ] Add optional "numbytes" arg to ns_conncptofp

[SourceForge.net <no...@so...>]

Feature Requests item #1156899, was opened at 2005-03-04 20:30
Message generated for change (Comment added) made by vasiljevic
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1156899&group_id=130646

Category: Tcl-API
Group: None
>Status: Closed
Resolution: Accepted
Priority: 5
Submitted By: Zoran Vasiljevic (vasiljevic)
Assigned to: Zoran Vasiljevic (vasiljevic)
Summary: Add optional "numbytes" arg to ns_conncptofp

Initial Comment:
The current ns_conncptofp will copy the content to the 
file. The entire content. What is missing is to tell to the 
command to copy only part of the content. 
The proposed API would be: 
 
 ns_conncptofp channel ?numbytes? 
 
The numbytes can be: 
 ommited (procedure does what it did before) 
  >0 (will copy somany bytes up to contentLength or  
  end-of-data whatever comes first) 
 0 (will do nothing) 
 -1 (will behave as when argument is ommited)  
 
This command is one of those which support the 
old-compat connId argument. So the actual API is 
as of today: 
 
  ns_conncptofp ?connId? channel 
 
I would vote to scrap the connId parsing. 

----------------------------------------------------------------------

>Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 09:08

Message:
Logged In: YES 
user_id=95086

Rewritten to leave contentLength intact and use reqPtr->avail. 

----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-18 08:50

Message:
Logged In: YES 
user_id=95086

Good point. Thanks for looking into this. I will now look into 
the reqPtr->avail instead of contentLength. 

----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-18 01:39

Message:
Logged In: YES 
user_id=87254


This change introduces a new side-effect:  the contentLength
is adjusted down by the number of bytes coppied.  Does this
need to happen?  The C API doesn't do this.

I think reqPtr->avail needs to be checked rather then
connPtr->contentLength.  reqPtr->next and reqPtr->avail will
be managed by the C API so no need to adjust them here. 

Also, as the Tcl API may return less than you asked for
rather than error out like the C API, should it return the
number of bytes copied?



----------------------------------------------------------------------

Comment By: Zoran Vasiljevic (vasiljevic)
Date: 2005-03-17 21:24

Message:
Logged In: YES 
user_id=95086

Added into CVS.

    ns_conncptofp ?-bytes tocopy? channel
 
The command will silently accept <=0 bytes and revert to
copying all that is there. Also, if more than available is
requested
it will silently copy all that is there. Otherwise, -bytes
bytes will 
be copied to the channel. Channel must be opened for writing
otherwise error is thrown.

Oh yes, Stephen, I love the arg parsing code This is really
a relief. After years any years of parsing... heh..


----------------------------------------------------------------------

Comment By: Stephen Deasey (sdeasey)
Date: 2005-03-05 00:30

Message:
Logged In: YES 
user_id=87254


This seems reasonable.

Should it look like this though?
    ns_conncptofp ?-bytes tocopy? channel

The reason it is attractive to drop the connid arg is that
with two optional args it becomes a pain to parse.  I'm also
thinking of how ns_sendmail has been extended over time by
appending optional args.

The C version of this command returns an error if you ask
for more bytes than are available.  Should the Tcl version
silently truncate?  Maybe it should return the number of
bytes actually copied?

Does it make sense to ask for 0 (zero) bytes?  Maybe bytes
<= 0 should revert to default behaviour which is to copy all
that's available?

The existing error message "could not copy content (likely
client disconnect)" is wrong.  This might have been the case
in 3.x, but not 4.x.



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=719009&aid=1156899&group_id=130646