naviserver-devel — Developer discussion

Re: [naviserver-devel] ns_parseargs example

[Zoran V. <zv...@ar...>]

Am 27.05.2005 um 11:34 schrieb Stephen Deasey:

> Where we left the discussion last time is how to deal with type
> checking.  Boolean switches and range values are basically just forms
> of type checking.

Hmm.... what I understood here was data-type checking as in
integer/string type of thing.

Under boolean support, I actually ment *absence* or *existence*
of the option. I do not know how I could achieve this now, unless
I say:

    ns_parseargs {{-boolean 0} args}

but this is also half-baked.

OK, lets think about type-checking. I would definitely say that
boolean type-checking (as described above) should be in.
The same for option-lists. Those two are most commonly found
and needed. Other type of checking would not be needed.
So, if I say:

     ns_parseargs {-a args}

then anything what is given to "-a" should be accepted as is.
Something like "string is ..." checking should be left to the
programmer.

I believe that underlying C-api already does the above, or?


>
> Depending on how type checking is implemented, then the syntax of an
> arg specification may change.  You may or may not end up pulling your
> hair out later if you change all your code to use ns_proc now...  :-)
>


Right. Lets fix this now. We can talk about the syntax when/if we
agree on the scope of the implemented type-checking (as above).
Deal?

Zoran 

Re: [naviserver-devel] ns_parseargs example

[Stephen D. <sd...@gm...>]

On 5/26/05, Zoran Vasiljevic <zv...@ar...> wrote:
>=20
> Ah. Good tip. What I was also looking is the usage of "?" in the
> specstring (never understood this one).


The '?' signifies an optional argument, but it's used in the C API
only.  Tcl already has defaults, which is effectively an optional
argument.  Having the programmer explicitly handle optional args when
using the C API is faster and more natural, I think.

All that's surfaced in the Tcl API so far is options and the end of
options marker.  Args, defaults and the 'args' list is also
implemented, but of course that's standard Tcl behaviour.

Where we left the discussion last time is how to deal with type
checking.  Boolean switches and range values are basically just forms
of type checking.

Depending on how type checking is implemented, then the syntax of an
arg specification may change.  You may or may not end up pulling your
hair out later if you change all your code to use ns_proc now...  :-)

Re: [naviserver-devel] ns_parseargs example

[Zoran V. <zv...@ar...>]

Am 26.05.2005 um 17:41 schrieb Stephen Deasey:

>
> There's some use of ns_parseargs in the nstest_http proc of the test
> harness, and of course in ns_parseargs.test.
>

Another one:

How would I give a list of acceptable values to an option?
For example:

    -a (1|2|3)

so the value of "a", if given, can be one of 1, 2 or 3 but not 4?

(Qurious) Zoran

Re: [naviserver-devel] ns_parseargs example

[Zoran V. <zv...@ar...>]

Am 26.05.2005 um 18:08 schrieb Zoran Vasiljevic:

>> proc ns_proc {name spec body} {
>>     proc $name {args} "
>>         ns_parseargs $spec \$args
>>         $body
>>     "
>> }
>>
>> ns_proc ex2 {-a {-b ""} -- x {y Y} args} {
>>     if {[info exists a]} {
>>         return [list $a $b $x $y $args]
>>     }
>>     return [list $b $x $y $args]
>> }
>

Ah.. you ment overriding "proc" itself?

So, instead of saying

   proc myproc {arg1 arg2 args} {
      ....
   }

one can also say:

   proc myproc {-a {-b ""} -- x {y Y} args} {
      ....
   }

So, in fact the "proc" becomes more clever?

Zoran

Re: [naviserver-devel] ns_parseargs example

[Zoran V. <zv...@ar...>]

Am 26.05.2005 um 17:41 schrieb Stephen Deasey:

> On 5/26/05, Zoran Vasiljevic <zv...@ar...> wrote:
>
>> Stephen,
>>
>> Can you please give us some examples how to use the
>> ns_parseargs call? I would like to integrate this
>> in our code now and would need to instruct all my
>> collegues how to use it.
>>
>> Just a couple of examples would be sufficient.
>>
>> Thanks
>> Zoran
>>
>
>
> proc ex1 {args} {
>     ns_parseargs {-a {-b ""} -- x {y Y} args} $args
>
>     if {[info exists a]} {
>         return [list $a $b $x $y $args]
>     }
>     return [list $b $x $y $args]
> }
>
>
> There's some use of ns_parseargs in the nstest_http proc of the test
> harness, and of course in ns_parseargs.test.

Ah. Good tip. What I was also looking is the usage of "?" in the
specstring (never understood this one).

So, the "-a" will collect whatever comes after "-a" argument in
the arguments list and set the "a" variable to the value. Clear.
If I omit "-a" from the argument list, I can use specstring

   {-a a.default.value}

and this will set "a" variable with "a.default.value".

Now, where does the "?" come?
Also, how do I specify:

   myproc -booleanarg myarg
   myproc myarg

I'd want to have variable booleanarg set to 1 if i have "-booleanarg"
in argument list and set to 0 if I dont. How to specify this
(if possible at all; I recall there was an issue with that)?

Sorry for all this questions but this lies quite a few weeks
since we last talked about it...

>
> We should probably define ns_proc to allow a more natural usage.  I
> didn't do that at first as we were discussing extra features such as
> type checking.  Do you think we can make that backwards compatible?
>
> proc ns_proc {name spec body} {
>     proc $name {args} "
>         ns_parseargs $spec \$args
>         $body
>     "
> }
>
> ns_proc ex2 {-a {-b ""} -- x {y Y} args} {
>     if {[info exists a]} {
>         return [list $a $b $x $y $args]
>     }
>     return [list $b $x $y $args]
> }
>

Hmmm... what do you mean by "backwards compatible"?
The ns_proc does not exist, as I see, hence there
is no compatibility to look after. We can of course
define the ns_proc anytime and it can coexist with
the Tcl proc. Did I understand you correctly at all?

Apropos type-checking, I would not go to that extent.
The number, position and default values of arguments
seem pretty fine and sufficient. I would not like to
make it more complicated than absolutely necessary.

Cheer's
Zoran

Re: [naviserver-devel] ns_parseargs example

[Stephen D. <sd...@gm...>]

On 5/26/05, Zoran Vasiljevic <zv...@ar...> wrote:
> Stephen,
>=20
> Can you please give us some examples how to use the
> ns_parseargs call? I would like to integrate this
> in our code now and would need to instruct all my
> collegues how to use it.
>=20
> Just a couple of examples would be sufficient.
>=20
> Thanks
> Zoran


proc ex1 {args} {
    ns_parseargs {-a {-b ""} -- x {y Y} args} $args

    if {[info exists a]} {
        return [list $a $b $x $y $args]
    }
    return [list $b $x $y $args]
}


There's some use of ns_parseargs in the nstest_http proc of the test
harness, and of course in ns_parseargs.test.

We should probably define ns_proc to allow a more natural usage.  I
didn't do that at first as we were discussing extra features such as
type checking.  Do you think we can make that backwards compatible?

proc ns_proc {name spec body} {
    proc $name {args} "
        ns_parseargs $spec \$args
        $body
    "
}

ns_proc ex2 {-a {-b ""} -- x {y Y} args} {
    if {[info exists a]} {
        return [list $a $b $x $y $args]
    }
    return [list $b $x $y $args]
}

[naviserver-devel] ns_parseargs example

[Zoran V. <zv...@ar...>]

Stephen,

Can you please give us some examples how to use the
ns_parseargs call? I would like to integrate this
in our code now and would need to instruct all my
collegues how to use it.

Just a couple of examples would be sufficient.

Thanks
Zoran

Re: [naviserver-devel] naming nsdb database drivers

[Zoran V. <zv...@ar...>]

Am 26.05.2005 um 04:05 schrieb Stephen Deasey:

> I'd like to import the nspostgres module from aolserver CVS as nsdbpg.
>  I have a new database driver loader I'm calling nsdbi, and one driver
> for it called nsdbipg.  There needs to be some way of distinguishing
> between the two types of drivers, and I think prefixing with either db
> or dbi should do it.
>
> There are already three driver modules in CVS, nsext and nspd which we
> split out from core, and Vlad's new nsfreetds.  I propose new names
> of: nsdbext, nsdbpd and nsdbtds.
>
> If no one objects, I'll get the SF admins to rename those directories
> and then I'll touch-up the Makefile's.
>
>

No objections.
Zoran

[naviserver-devel] naming nsdb database drivers

[Stephen D. <sd...@gm...>]

I'd like to import the nspostgres module from aolserver CVS as nsdbpg.
 I have a new database driver loader I'm calling nsdbi, and one driver
for it called nsdbipg.  There needs to be some way of distinguishing
between the two types of drivers, and I think prefixing with either db
or dbi should do it.

There are already three driver modules in CVS, nsext and nspd which we
split out from core, and Vlad's new nsfreetds.  I propose new names
of: nsdbext, nsdbpd and nsdbtds.

If no one objects, I'll get the SF admins to rename those directories
and then I'll touch-up the Makefile's.

Re: [naviserver-devel] Patches

[Zoran V. <zv...@ar...>]

Am 20.05.2005 um 20:18 schrieb Vlad Seryakov:

> Hello guys,
>
> Looks like some of my posts get lost, i do not see tham in the SF  
> atchives.
>
> My question is, if nobody object i want to commit latest UDP patch  
> and add nsudp module to modules section.
>
> Also, i have a bunch of modules that i commited into AS, i would  
> like to commit them into modules for NS as well.
>

Go ahead.
Zoran

[naviserver-devel] Patches

[Vlad S. <vl...@cr...>]

Hello guys,

Looks like some of my posts get lost, i do not see tham in the SF atchives.

My question is, if nobody object i want to commit latest UDP patch and 
add nsudp module to modules section.

Also, i have a bunch of modules that i commited into AS, i would like to 
commit them into modules for NS as well.

-- 
Vlad Seryakov
571 262-8608 office
vl...@cr...
http://www.crystalballinc.com/vlad/

Re: [naviserver-devel] ns_urlencode compat

[Zoran V. <zv...@ar...>]

Am 11.05.2005 um 14:12 schrieb Stephen Deasey:

>
> You'll have a syncronization problem: you'll have to throw the switch
> on all servers at the same time.
>

Imagine: there are about 500 installations world-wide we have currently.
So, this one is not an option...

> Another alternative might be to patch the AOLserver instances to
> always encode spaces as %20, which is always valid.  Then you only
> need to ensure that you don't install naviserver on a network with an
> unpatched AOLserver.

Again not an option...

>
> Oh, I guess you'd still have to patch naviserver to also encode using
> %20, but at least that's a valid encoding.
>

Ehm... what I found out is that the "+" sign should be "accepted" by
the Naviserver as one of the valid escaped ' ' chars. So, I have
added URLDECODE_RELAXED (undefined by default) in the UrlDecode() so
the things encoded as '+'s will be converted to ' 's even when walking
the *path* part of the url (because this is how the aolserver would
encode the string). Now, I understand this ain't very clean but
it help us for now. I will remove this define when we complete the
transition at the customer sites. Anyways, it should not hurt anybody
I believe.

Zoran

Re: [naviserver-devel] ns_urlencode compat

[Stephen D. <sd...@gm...>]

On 5/10/05, Zoran Vasiljevic <zv...@ar...> wrote:
> Hi,
>=20
> Eh... seems that we do still have some (compat) problems there...
>=20
> I do not doubt that the current implementation in urlencode.c
> is OK, but we do have many existing installations running on the
> AOLserver and a couple of new ones running with NaviServer now.
>=20
> There are subtle differences in way the urldecoder works.
>=20
> The naviserver one will not expand "a+b" into "a b" if walking
> over the path part. This seems to be correct. But the AOLserver
> urlencoder encodes the " " in both url path part *and* the query
> part with "+". Therefore we now have a problem when operating between
> the Naviserver and Aolserver instances.
>=20
> We are now stucked.
>=20
> Question: is there an easy way to #ifdef parts of the code so
> the encoding/decoding machinery becomes compatible to the aolserver one?
> I will need those compat switches until we get all our customers
> (and they all their instances) updated. This can take some considerable
> time, though, perhaps a year or so.
>=20
> I have been walking over the sources and there seems to be quite
> a few changes in this respect in various parts of the code.
> In order to save some (electronic archeology) time I thought I'd rather
> ask here. Stephen, do you have some idea how I can achieve this?


You'll have a syncronization problem: you'll have to throw the switch
on all servers at the same time.

Another alternative might be to patch the AOLserver instances to
always encode spaces as %20, which is always valid.  Then you only
need to ensure that you don't install naviserver on a network with an
unpatched AOLserver.

Oh, I guess you'd still have to patch naviserver to also encode using
%20, but at least that's a valid encoding.

Life's a beach   :-)

Re: [naviserver-devel] release ?

[Bernd E. <eid...@we...>]

Hi Stephen,

> I hate to say so at this late stage, but would it be possible to keep
> everything on Source Forge?  

:-) I would suggest to let Source Forge do the stuff you mentioned, the hard 
work of managing mirrored downloads, cvs, bug tracking etc. 

>The goal is to turn newbies into dedicated followers. 
>Therefore, the front page is for impressing newbies.

Right. But an impressing frontpage is no matter of the underlying technology.
It should be something different, more flexible, more impressing than what 
exists so far for the servers direct competitor ;-)

The website is soon to be finished, let's see how to improve it at that time. 
(Btw: if it will turn out to be unstable, unnecessary etc. then switch back 
to SF, of course!).

Regards,
Bernd.

Re: [naviserver-devel] release ?

[Stephen D. <sd...@gm...>]

On 5/5/05, Bernd Eidenschink <eid...@we...> wrote:
> > Am 04.05.2005 um 16:45 schrieb Vlad Seryakov:
> > > We just do not have Web site, at least first page with some details
> > > would be nice. I guess, hosting on sourceforge gives us html version
> > > of web page only.
> >
> > Bernd can take this on his shoulders, I think. Bernd?
>=20
> Yes. I will set up naviserver the next days on a server already intended =
for
> the website. We could use www.servercult.com/naviserver or smth. like tha=
t as
> we have this domain already. But this is not so important as we can chang=
e it
> to whatever whenever we want.
>=20
> I'll keep you up to date. What texts do you think we need for the start (=
the
> first promotion)?
>=20
> - Vision/Mission
> - Background Info
> - Where to download/How to checkout
> - development guideline(s)
> - ...
>=20
> Bernd.


Hi!

I hate to say so at this late stage, but would it be possible to keep
everything on Source Forge?  The nice thing about SF is that it's all
managed for us.  If a hard drive dies, we won't even notice.  Any one
of us can get busy/knocked over by a bus, without affecting any one
else.  We all allready have logins, and it's easy to add more, etc...

Vlad raises a good point: is SF limeted to static files only?  We get
PHP out of the box, a MySQL database, and a few hundred megs of
storage.  I believe they're open to adding more stuff if we need it.

Regarding the content, at this early stage we don't have a lot.  In my
ideal world we'd have a very focused front page with a message of why
naviserver is interesting.  Then we'd have a simple template we could
pour the other content into as we generate it.  I'm thinking about
something like these:

http://www.newsfirerss.com/
http://www.gnome.org/projects/f-spot/

It's blog-template-like in style with a big splashy graphic, but it's
extremely direct: here's the product (often a screen shot), here's a
big ol' download button, here's a tag line and a description, we're at
version X.  Things like forum postings and press releases are not
bubbled up to the front page.

The idea is there are two groups of people you're interested in: those
who don't know who you are, and dedicated followers.  Your dedicated
followers will be subscribed to the mailing list and watch the bug
tracker.  The goal is to turn newbies into dedicated followers.=20
Therefore, the front page is for impressing newbies.

Besides, designing the above is much more fun than designing a
portal-like page ;-)


Just MHO...

Re: [naviserver-devel] release ?

[Stephen D. <sd...@gm...>]

Hi.  There's a couple of things I've been working on which I though
would be good additions for a first release, but I ran into a few
problems and I haven't time to get them working.  So if y'all are
happy to cut a tarball now, that's fine with me.

Re the numbering, I thought 4.99.0 suggested we're making pre-releases
towards 5.0.0.  How about we use even/odd unstable/stable releases?=20
i.e. when 5.0 is done we move to 5.1 immediately, make a number of
5.1.x releases, and when it's stable release 5.2.0.

We need to tag and branch.  We'd tag odd releases, branch even ones.=20
Odd releases are test releases, no promises.  Even releases are stable
releases, we may decide to backport bug fixes etc.  We tag
naviserver_4_99_0, we branch naviserver_5_0_0, we tag
naviserver_5_0_1, we tag naviserver_5_1_13, we branch naviserver_5_2_0
...

~

One thing it would be nice to have is a 'make dist' target.  You
really want the version number embedded in the configure script for
that.  Then you want to derive the version info in the header from
that same number, rather than rely on manual syncronization.  Which is
a great oppertunity to introduce a config.h header and revamp the
build system, including standard install paths.  But then you need to
teach the server init system how to find and load modules, and that
code is completely broken.

Hey, I got carried away  :-)

I'll try and break some of that up into sensible chunks early on for 4.99.1=
.


On 5/4/05, Zoran Vasiljevic <zv...@ar...> wrote:
> Hi friends,
>=20
> I have updated our product to use NaviServer now. It will be
> distributed as
> update to about 500 installations worldwide.
> I have tested it thoroughly and apart from the (already noted)
> ns_urlencode
> (incompatibility) it seems that other things are backward-compatible
> with the
> aolserver. We still do not use any new features, though. This will be
> added
> as we pass first field-tests.
>=20
> I could use this occasion to raise the "are we about to make a release"
> question
> again. As it seems, it would be fine for us to tag the CVS and make a
> tarball.
> What do you think? If yes, which version? 4.99?
>=20
> Zoran

[naviserver-devel] ns_urlencode compat

[Zoran V. <zv...@ar...>]

Hi,

Eh... seems that we do still have some (compat) problems there...

I do not doubt that the current implementation in urlencode.c
is OK, but we do have many existing installations running on the
AOLserver and a couple of new ones running with NaviServer now.

There are subtle differences in way the urldecoder works.

The naviserver one will not expand "a+b" into "a b" if walking
over the path part. This seems to be correct. But the AOLserver
urlencoder encodes the " " in both url path part *and* the query
part with "+". Therefore we now have a problem when operating between
the Naviserver and Aolserver instances.

We are now stucked.

Question: is there an easy way to #ifdef parts of the code so
the encoding/decoding machinery becomes compatible to the aolserver one?
I will need those compat switches until we get all our customers
(and they all their instances) updated. This can take some considerable
time, though, perhaps a year or so.

I have been walking over the sources and there seems to be quite
a few changes in this respect in various parts of the code.
In order to save some (electronic archeology) time I thought I'd rather
ask here. Stephen, do you have some idea how I can achieve this?

Thanks,
Zoran

  

Re: [naviserver-devel] release ?

[Zoran V. <zv...@ar...>]

Am 10.05.2005 um 14:49 schrieb Bernd Eidenschink:

> ok. i'll try to set it up today.

No rush... keep it easy going :)
Zoran

Re: [naviserver-devel] release ?

[Bernd E. <eid...@we...>]

> I believe we can omit all those for the beginning except the "here is
> the
> download link to the last release" type of thing. It costs (much) time
> to get
> it all written (and maintained) and we are all pretty busy...

ok. i'll try to set it up today.

Re: [naviserver-devel] release ?

[Zoran V. <zv...@ar...>]

Am 05.05.2005 um 16:49 schrieb Bernd Eidenschink:
>
> Yes. I will set up naviserver the next days on a server already 
> intended for
> the website. We could use www.servercult.com/naviserver or smth. like 
> that as
> we have this domain already. But this is not so important as we can 
> change it
> to whatever whenever we want.
>
> I'll keep you up to date. What texts do you think we need for the 
> start (the
> first promotion)?
>
> - Vision/Mission
> - Background Info
> - Where to download/How to checkout
> - development guideline(s)
> - ...

I believe we can omit all those for the beginning except the "here is 
the
download link to the last release" type of thing. It costs (much) time 
to get
it all written (and maintained) and we are all pretty busy...

Zoran

Re: [naviserver-devel] release ?

[Bernd E. <eid...@we...>]

> Am 04.05.2005 um 16:45 schrieb Vlad Seryakov:
> > We just do not have Web site, at least first page with some details
> > would be nice. I guess, hosting on sourceforge gives us html version
> > of web page only.
>
> Bernd can take this on his shoulders, I think. Bernd?

Yes. I will set up naviserver the next days on a server already intended for 
the website. We could use www.servercult.com/naviserver or smth. like that as 
we have this domain already. But this is not so important as we can change it 
to whatever whenever we want.

I'll keep you up to date. What texts do you think we need for the start (the 
first promotion)?

- Vision/Mission
- Background Info
- Where to download/How to checkout
- development guideline(s)
- ...

Bernd.

Re: [naviserver-devel] release ?

[Zoran V. <zv...@ar...>]

Am 04.05.2005 um 16:54 schrieb Vlad Seryakov:

>> Any ideas what? Stephen suggested 4.99. Why not use that? Or more=20
>> radically, 5.0 as we
>> do have virtual hosting now which can justify the release bump.
>
> Okay, 4.99 and we'll switch to 5.0 when docs will be ready.
>

Good. =10I'm still having open discussion with Andreas (Kupries) from =
Tcl=20
project
about doctools package we could use to write nice-looking docs. I think=20=

I
will have to speed-up this.

Zoran

Re: [naviserver-devel] release ?

[Vlad S. <vl...@cr...>]

> 
> Any ideas what? Stephen suggested 4.99. Why not use that? Or more 
> radically, 5.0 as we
> do have virtual hosting now which can justify the release bump.

Okay, 4.99 and we'll switch to 5.0 when docs will be ready.

-- 
Vlad Seryakov
571 262-8608 office
vl...@cr...
http://www.crystalballinc.com/vlad/

Re: [naviserver-devel] release ?

[Zoran V. <zv...@ar...>]

Am 04.05.2005 um 16:45 schrieb Vlad Seryakov:
>
> We just do not have Web site, at least first page with some details 
> would be nice. I guess, hosting on sourceforge gives us html version 
> of web page only.
>

Bernd can take this on his shoulders, I think. Bernd?

> Also, until we have docs we can release snapshots without tagging CVS, 
> kind of working release, i am not comfortable with CVS tagging and 
> branching:-)))
>

Oh, tagging is absolutely crucial! We do not need to work on the 
branches (yet) but it
is important for us to be able to checkout the released version and 
make any fixes
if necessary. You could ignore the branches if you like and always 
commit to cvs-head
but we must have versioning.

> Using versions 4.0.x will be in collision with AS, i think new version 
> scheme should be used.
>

Any ideas what? Stephen suggested 4.99. Why not use that? Or more 
radically, 5.0 as we
do have virtual hosting now which can justify the release bump.

Zoran

Re: [naviserver-devel] release ?

[Vlad S. <vl...@cr...>]

I agree we can make public release, we wanted to do this with docs but 
it may take longer, so public release with official use in commercial 
product could be enough reason to make first appearance.

We just do not have Web site, at least first page with some details 
would be nice. I guess, hosting on sourceforge gives us html version of 
web page only.

Also, until we have docs we can release snapshots without tagging CVS, 
kind of working release, i am not comfortable with CVS tagging and 
branching:-)))

Using versions 4.0.x will be in collision with AS, i think new version 
scheme should be used.

Zoran Vasiljevic wrote:
> Hi friends,
> 
> I have updated our product to use NaviServer now. It will be distributed as
> update to about 500 installations worldwide.
> I have tested it thoroughly and apart from the (already noted) ns_urlencode
> (incompatibility) it seems that other things are backward-compatible 
> with the
> aolserver. We still do not use any new features, though. This will be added
> as we pass first field-tests.
> 
> I could use this occasion to raise the "are we about to make a release" 
> question
> again. As it seems, it would be fine for us to tag the CVS and make a 
> tarball.
> What do you think? If yes, which version? 4.99?
> 
> Zoran
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: NEC IT Guy Games.
> Get your fingers limbered up and give it your best shot. 4 great events, 4
> opportunities to win big! Highest score wins.NEC IT Guy Games. Play to
> win an NEC 61 plasma display. Visit http://www.necitguy.com/?r=20
> _______________________________________________
> naviserver-devel mailing list
> nav...@li...
> https://lists.sourceforge.net/lists/listinfo/naviserver-devel

-- 
Vlad Seryakov
571 262-8608 office
vl...@cr...
http://www.crystalballinc.com/vlad/