mod-pubsub-developer — Developer list for mod_pubsub

[Mod-pubsub-developer] examples for mod_pubsub usage?

[<mod...@li...>]

Hello,
 	Thanks for creating such an interesting (and potentially useful :) 
piece of software!
 	Are there any examples available demonstrating the usage of 
mod_pubsub with javascript clients?  Thanks!

 	J. Patrick Bedell
 	jp...@in...

Re: [Mod-pubsub-developer] Re: running mod_pubsub_2004-04-20 with mod_perl

[<mod...@li...>]

Hey Bob,

So we aren't "officially" maintaining that particular client and server any
more.  Some of the devs on this list might be doing something with it --
Kragen? -- but not me.

  -- Joyce


--- mod...@li... wrote:

> I can answer portions of my own questions at this point.  If others 
> have tried to run the old Perl server recently, perhaps they're running 
> into the same problems.  I should have mentioned that my OS is FreeBSD 
> 5.3, since that could be relevant for the libraries used.  I have Perl 
> 5.8.6 installed.
> 
> My confusion about clients and servers was cleared up by looking at the 
> code for pubsub_test.cgi.  It mentioned that you can run it at the 
> command line (as you can any script that uses CGI.pm).  You do need to 
> set some environmental variables to feed it the arguments it would get 
> when called from a browser.  In any case this lets you test the server 
> setup by running scripts on the server, so you're not dependent on any 
> browser details.
> 
> After some experimentation I found that most of the tests were failing 
> because PubSub::Server was failing to flush its output buffers.  There 
> is a 300 second delay built into the code, so if you sit around long 
> enough, you get to see the test fail.  If you're less patient, you 
> think it's hung.  I added commands to flush the output buffers after 
> three "send" commands in Server.pm, after which the majority of the 
> tests gave an "OK" report.  So far I still don't understand why the `$| 
> = 1;' command in PubSub::Server->route_to_javascript doesn't do the 
> trick.  I had presumed it would force unbuffered output, and I haven't 
> found anything in the Apache 2 or mod_perl documentation to suggest 
> anything to the contrary.
> 
> With these patches all tests but the last two -- replay of events and 
> replay of events with warp factor -- are now OK.  I have yet to explore 
> the nature of these failures.  Are they something I should worry about?
> 
> If there is an interest in updating the software or the INSTALL 
> document, please let me know, and I can attempt a slightly more 
> coherent description of what I've found.
> 
> Thanks,
> Bob Carlitz
> 
> 
> On May 7, 2005, at 1:58 PM, 
> mod...@li... wrote:
> 
> > I was interested in using the Perl server from the 2004-04-20 
> > distribution, assuming it can be integrated with other files on the 
> > server.  I'm running Apache 2.1.3 with mod_perl 1.999_20 but was 
> > unable to get things running without some adjustments, as detailed 
> > below:
> >
> > The configuration generated by ModPubSub.pm failed with an error 
> > message "Option ExecCGI now allowed here." I copied the configuration 
> > into httpd.conf and commented out the "$Directory =" and "$Location =" 
> > lines in ModPubSub.pm. I updated the syntax to use 
> > "PerlResponseHandler ModPerl::Registry" and "PerlOptions 
> > +ParseHeaders". I also created an alias for /kn to 
> > /var/www/mod_pubsub/cgi-bin/pubsub.cgi. This lets me start up Apache 
> > with mod_pubsub running.
> >
> > Most of the test programs failed after the one that tries to create a 
> > tunnel.  I'm using a Mac and tried Safari and Firefox with the same 
> > results.  I did verify that mod_pubsub was generating directories and 
> > files in /kn_events.  Should I conclude that my problems are on the 
> > browser side rather than the server side?
> >
> > If I can get this stuff running, I could perhaps help with your 
> > documentation effort.  I'm not sure how much information you had on 
> > the currently unavailable Wiki, but I'm finding things pretty hard to 
> > dig through.
> >
> > Thanks,
> > Bob Carlitz
> >
> >
> >
> > -------------------------------------------------------
> > 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
> > _______________________________________________
> > Mod-pubsub-developer mailing list
> > Mod...@li...
> > https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> 
> 
> 
> -------------------------------------------------------
> 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
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> 

[Mod-pubsub-developer] Re: running mod_pubsub_2004-04-20 with mod_perl

[<mod...@li...>]

I can answer portions of my own questions at this point.  If others 
have tried to run the old Perl server recently, perhaps they're running 
into the same problems.  I should have mentioned that my OS is FreeBSD 
5.3, since that could be relevant for the libraries used.  I have Perl 
5.8.6 installed.

My confusion about clients and servers was cleared up by looking at the 
code for pubsub_test.cgi.  It mentioned that you can run it at the 
command line (as you can any script that uses CGI.pm).  You do need to 
set some environmental variables to feed it the arguments it would get 
when called from a browser.  In any case this lets you test the server 
setup by running scripts on the server, so you're not dependent on any 
browser details.

After some experimentation I found that most of the tests were failing 
because PubSub::Server was failing to flush its output buffers.  There 
is a 300 second delay built into the code, so if you sit around long 
enough, you get to see the test fail.  If you're less patient, you 
think it's hung.  I added commands to flush the output buffers after 
three "send" commands in Server.pm, after which the majority of the 
tests gave an "OK" report.  So far I still don't understand why the `$| 
= 1;' command in PubSub::Server->route_to_javascript doesn't do the 
trick.  I had presumed it would force unbuffered output, and I haven't 
found anything in the Apache 2 or mod_perl documentation to suggest 
anything to the contrary.

With these patches all tests but the last two -- replay of events and 
replay of events with warp factor -- are now OK.  I have yet to explore 
the nature of these failures.  Are they something I should worry about?

If there is an interest in updating the software or the INSTALL 
document, please let me know, and I can attempt a slightly more 
coherent description of what I've found.

Thanks,
Bob Carlitz


On May 7, 2005, at 1:58 PM, 
mod...@li... wrote:

> I was interested in using the Perl server from the 2004-04-20 
> distribution, assuming it can be integrated with other files on the 
> server.  I'm running Apache 2.1.3 with mod_perl 1.999_20 but was 
> unable to get things running without some adjustments, as detailed 
> below:
>
> The configuration generated by ModPubSub.pm failed with an error 
> message "Option ExecCGI now allowed here." I copied the configuration 
> into httpd.conf and commented out the "$Directory =" and "$Location =" 
> lines in ModPubSub.pm. I updated the syntax to use 
> "PerlResponseHandler ModPerl::Registry" and "PerlOptions 
> +ParseHeaders". I also created an alias for /kn to 
> /var/www/mod_pubsub/cgi-bin/pubsub.cgi. This lets me start up Apache 
> with mod_pubsub running.
>
> Most of the test programs failed after the one that tries to create a 
> tunnel.  I'm using a Mac and tried Safari and Firefox with the same 
> results.  I did verify that mod_pubsub was generating directories and 
> files in /kn_events.  Should I conclude that my problems are on the 
> browser side rather than the server side?
>
> If I can get this stuff running, I could perhaps help with your 
> documentation effort.  I'm not sure how much information you had on 
> the currently unavailable Wiki, but I'm finding things pretty hard to 
> dig through.
>
> Thanks,
> Bob Carlitz
>
>
>
> -------------------------------------------------------
> 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
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer

[Mod-pubsub-developer] running mod_pubsub_2004-04-20 with mod_perl

[<mod...@li...>]

I was interested in using the Perl server from the 2004-04-20 
distribution, assuming it can be integrated with other files on the 
server.  I'm running Apache 2.1.3 with mod_perl 1.999_20 but was unable 
to get things running without some adjustments, as detailed below:

The configuration generated by ModPubSub.pm failed with an error 
message "Option ExecCGI now allowed here." I copied the configuration 
into httpd.conf and commented out the "$Directory =" and "$Location =" 
lines in ModPubSub.pm. I updated the syntax to use "PerlResponseHandler 
ModPerl::Registry" and "PerlOptions +ParseHeaders". I also created an 
alias for /kn to /var/www/mod_pubsub/cgi-bin/pubsub.cgi. This lets me 
start up Apache with mod_pubsub running.

Most of the test programs failed after the one that tries to create a 
tunnel.  I'm using a Mac and tried Safari and Firefox with the same 
results.  I did verify that mod_pubsub was generating directories and 
files in /kn_events.  Should I conclude that my problems are on the 
browser side rather than the server side?

If I can get this stuff running, I could perhaps help with your 
documentation effort.  I'm not sure how much information you had on the 
currently unavailable Wiki, but I'm finding things pretty hard to dig 
through.

Thanks,
Bob Carlitz

[Mod-pubsub-developer] More about the mod-pubsub server rewrite...

[<mod...@li...>]

Some more notes...


1. It will incorporate Twisted 2.0 as well as the architectural
changes we envision, and will fix all memory leaks. It should fly and
scale just fine.


2. Based on feedback, we decided there's no real point in changing the
protocol in major ways.  The big architectural changes involve moving
ancillary functionality out to hooks -- like persistence and replay
and auth -- but those things have never worked properly in
Mod-pubsub before now anyway.


3. We will provide beta drops in early May to any company that
contributes to paying for the development. Please see me if you're
interested.


4. One of our users wrote:

> We rely on ordered delivery for our system. Ideally, we would like our
> system to handle each published event. But sometimes our publishers
> publish more data than we can process. Our solution has been to
> have the listener publish a "self-addressed stamped envelope" on
> the topic. The subscriber listens on the topic until the "self-addressed
> stamped envelope" is received by the listener. The listener keeps the
> most recently received payload about a particular network node.
> We would have to come up with some other solution if we could
> not rely on ordering.

To this we respond: Apparently can keep ordering -- turns out to be
less expensive to implement than we thought.


5. One of our users wrote:

> I'd like to explore a use case to see if the new server architecture might
> be able to support it. 
>  
> We have a large number of devices in the field. We need to know when a
> device goes down or loses network connectivity.  We currently do this using
> an HTTP heartbeat mechanism where the device calls a heartbeat page using
> HTTP GET on a regular interval and passes in its device id.  If a device
> doesn't check in on the expected interval (or 2) then we suspect that the
> device is down or has lost network connectivity.  This is a simple mechanism
> and works fine, but is a headache when you need to scale up to lots of
> devices. You end up scaling up the server farm or stretching out the
> heartbeat interval (or both). 
>  
> What if we could use the pubsub server to manage connectivity monitoring... 
> We are going to have a connection to the pubsub server from every device
> anyway. If we could write ConnectionOpen and ConnectionClose hooks
> to notice when devices come and go then we could piggyback on
> mod-pubsub (or repubsub or rerepubsub or whatever it is going to be
> called ;-) to get our connectivity monitoring for free. On the surface,
> based on what I know, it seems reasonable; the only possible catch
> I can see is how do we identify the particular device/client on
> connection open and close. Journal path? 
>  
> What do you think? Should this be doable with the new server?   

To which we responded:The ConnectionOpen and ConnectionClose
hooks are already in the plan.

As for identifying the particular device... you can't just send the
device ID?

Actually, this is a brilliant use-case for Mod-pubsub and would
be helpful to a lot of people.  You can imagine the cool things
you could do... realtime monitoring, but also easily sending an
SMS or email or IM to notify people that a connection had
gone down.  W00t!

To which our user responded:

> I could... and that would work on connect, but not on disconnect.
> If a device is crashing or suddenly loses network connectivity, it
> doesn't have a chance to send a message. We need to somehow
> associate a connection with the device id so that when the
> connection breaks we know which device it was. It seems that
> passing a device id with the connection establishment would be
> the cleanest and most symmetrical. When the connection is
> opened, we could make the association between the device id
> and the connection. When the connection closes we would need
> to identify which connection closed and determine which device
> was associated with that connection. The trick is somehow
> injecting the device id in the tunnel establishment. 

To which we respond: it will be done, as requested, as part of the new
architecture. This just keeps getting better...

Can I get another amen,
  Adam


----
ifindkarma (ahem) at gmail yadda yadda yadda dot booya com w00t w00t 

Gary, this is a dangerous mission. If you happen to get captured,
suicide may be the more humane option. Here take this.
[hands Gary a hammer] 
   -- Team America: World Police

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Adam:

Is your information current?  I was under the impression that a 
streaming design was now in place for Active MQ and/or xmlBlaster's HTTP 
capabilities.  But that's probably wrong, I am vaguely remembering a 
feeling I got while skimming some docs a while back.

:-)

Either/both of these projects support dynamic topic creation and some of 
the core requirements to achieve REST client access characteristics.

I think mod-pubsub should consider layering itself onto these two quite 
popular, and quite active, MOM projecs.  Think "REST pubsub module" for 
MOM instead of for Apache.  The client-access characteristics that would 
emerge could act as a motivator for the MOM community to adopt REST 
architectural principles.  The current mod-pubsub server could/should 
interact well with these modules, allowing it to become a "subsidiary" 
or "proxy" event notification server, giving developers the choice of 
using the lightweight mod-pubsub solution in contexts where the 
full-blown MOM solution is inappropriate, yet connecting the two, or 
switching to the MOM solution, without ever changing APIs.

Alright, I'm just blowing hot air, but it was fun.

A.

mod...@li... wrote:

>A. wrote:
>
>  
>
>>>In the interest of strengthening everyone's knowledge base, what did you
>>>compare against and what were your concerns?  I've noticed that
>>>xmlBlaster and Activ MQ both offer HTTP oriented methodologies for
>>>asynchronous notification.  But I am not sure how robust they are, nor
>>>how native/central to the design they are, nor how well they work in
>>>javascript contexts.
>>>      
>>>
>
>  
>

Re: [Mod-pubsub-developer] Re: Quick question...

[<mod...@li...>]

RE: [Mod-pubsub-developer] Re: Quick question...Docs are in the =
distribution (apparently not directly linked on mod-pubsub.org)
http://mod-pubsub.org/downloads.html

Also here in CVS
http://cvs.sourceforge.net/viewcvs.py/mod-pubsub/mod_pubsub/kn_docs/

Protocol guide (very detailed, thanks Ben!)
http://cvs.sourceforge.net/viewcvs.py/*checkout*/mod-pubsub/mod_pubsub/kn=
_docs/pubsub_protocol.html?rev=3D1.9

Developers guide (summary and javascript client API)
http://cvs.sourceforge.net/viewcvs.py/*checkout*/mod-pubsub/mod_pubsub/kn=
_docs/developer_guide.html?rev=3D1.7


  ----- Original Message -----=20
  From: mod...@li...=20
  To: Joyce Park ; mod...@li... ; Adam =
Rifkin=20
  Sent: Sunday, April 03, 2005 11:30 PM
  Subject: RE: [Mod-pubsub-developer] Re: Quick question...




  > There is an sourceforge project called 'destiny' that is an HTTP=20
  > bridge to JMS.=20

  Excellent suggestion.=20

  > You could use that as a starting point to implement that same HTTP =
API=20
  > used by mod-pubsub.=20

  Were can I find (either files in distribution or docs) details on the =
HTTP API?=20
  Is there any info on the mechanism used by mod-pubsub for the keep =
alive connection?=20
  I am assume this is accomplished by both, having 'keep-alive' plus =
other info in the header?=20

  Oliver Muoto=20
  ol...@mu...=20

  ------------------------------------------------------- SF email is =
sponsored by - The IT Product Guide Read honest & candid reviews on =
hundreds of IT Products from real users. Discover which products truly =
live up to the hype. Start reading now. =
http://ads.osdn.com/?ad_ide95&alloc_id=14396&op=3Dick =
_______________________________________________ Mod-pubsub-developer =
mailing list Mod...@li... =
https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer 

Re: [Mod-pubsub-developer] Re: Quick question...

[<mod...@li...>]

> Uh... this email wasn't signed, but I think it's Dierken?  If so, hi Mike!
Yep. Howdy!

>
> Just to clarify, the Mod-pubsub SERVER API has not changed at all yet,
except
> for one tiny thing that was necessary for the Flash client but should not
> impact anything else.  The JavaScript API is the thing that changed
> significantly.  JP
Oh. My mistake. The java clients in CVS should work then. I think both parse
the text format that is provided by the server and puts the data into a
Hashmap and calls 'onMessage()' for the registered listeners (at least in my
version).



>
>
> --- mod...@li... wrote:
>
> > There is an sourceforge project called 'destiny' that is an HTTP bridge
to
> > JMS.
> > You could use that as a starting point to implement that same HTTP API
used
> > by mod-pubsub. There's nothing magic about the server implementation of
> > mod-pubsub, it's all in the protocol. If mod-pubsub had a multi-server
> > protocol for a constellation of messaging servers, that would be more
work
> > to 'port', but still doable.
> >
> > The java client I built is on the old mod-pubsub API, I don't think it's
> > compatible with the new server implementation.
> > There are actually two java client libraries in CVS for mod-pubsub. One
has
> > lots of unit tests. Mine doesn't... but still works.
> >
> > Let me know if you need advice or additional work for a java client or
java
> > server - i'd love to help implement a java server.
> >
> > I'd also like to implement an RSS format for 'change-lists' (lists of
recent
> > events) that use 'next' and 'prev' links to traverse past and future
> > history - I think that's the correct approach for a REST based event
> > notification system.
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
>

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

A:

> xmlBlaster and Activ MQ's HTTP for asynchronous notification either
> require polling or a proxy that can accept connections across the
> firewall. The beauty (and scariness) of mod-pubsub's approach is the
> use of streaming HTTP to keep connections persistent so no polling and
> no proxy needed.


At what level of granularity is polling not required? The reason I ask is that
the Python pubsublib client very much uses polling:

    while 1:
        asyncore.poll(scheduler.timeout())
        # print "\n\n asyncore.poll \n"
        scheduler.run()

Going back to the original question I had, this is the core of the problem since
the while loop is very much blocking and is processor intensive. From the
original Medusa documentation on asyncore, I offer the following quote:

"Now that you know what select() does, you're ready for the final piece of the
puzzle: the main polling loop. This is nothing more than a simple while loop
that continually calls select() with a timeout (I usually use a 30-second
timeout). Such a program will use virtually no CPU if your server is idle; it
spends most of its time letting the operating system do the waiting for it.
This is much more efficient than a busy-wait loop."

Hence the solution was to do just this: replace the while loop with the loop()
command which is much more CPU-friendly.

  asyncore.loop()

Anyone interested in embedding the client in a larger python project will likely
need to do this. Ultimately, this is a more efficient implementation and should
probably be merged into the code base. Do you want a patch?

Brian

-- 
  ====
brian lee dae yung
biomimetic art and research
  mux space . com

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Adam wrote:
> 
> By the way, I'm not sure anyone is currently using the Python client
> library in a project... though I'm sure one day someone will...
> 

I use the Python client in a small project to send out modification 
notifications from a Subversion-based xml document repository. Both js and java 
clients are subscribed to these messages and it's been in production for well 
over a year (probably closer to two) without missing a beat.

Robert

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

A. wrote:

> > In the interest of strengthening everyone's knowledge base, what did you
> > compare against and what were your concerns?  I've noticed that
> > xmlBlaster and Activ MQ both offer HTTP oriented methodologies for
> > asynchronous notification.  But I am not sure how robust they are, nor
> > how native/central to the design they are, nor how well they work in
> > javascript contexts.

xmlBlaster and Activ MQ's HTTP for asynchronous notification either
require polling or a proxy that can accept connections across the
firewall. The beauty (and scariness) of mod-pubsub's approach is the
use of streaming HTTP to keep connections persistent so no polling and
no proxy needed.

I'm certain neither of those projects work well in JavaScript
contexts. It's not what they were designed for, and it's the reason we
didn't use them in the first place.


Brian wrote:

> My basic criteria was finding a message-based system that was being actively
> developed/supported with Python bindings. I looked at a number of open source
> JMS implementations, most of which had no Python bindings. Activ MQ and Mom4J
> were the two that had support for Python, but Activ MQ seems too
> Windows-centric for serious use on Linux. Mom4J looked more promising, but
> their implementation was incomplete and their user community was like a ghost
> town. There's really no point in investing time in a dead project (unless it
> has killer technology). When I discovered mod_pubsub and the REST-based systems
> (been on my radar for a long time, but never delved into it) I was happy to
> find out that they overcome the confinement limitations of JMS. At this point,
> I'm ignorant of the security ramifications that this presents, which I'd like
> to find out more. FWIW, Mom4J provided hooks for authentication, but no actual
> implementation was present.
> 
> I'll probably be looking at some other REST systems, and I'll let you know the
> outcome.

Thanks, much appreciated.

By the way, I'm not sure anyone is currently using the Python client
library in a project... though I'm sure one day someone will...

   Adam

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Adam:

My basic criteria was finding a message-based system that was being actively
developed/supported with Python bindings. I looked at a number of open source
JMS implementations, most of which had no Python bindings. Activ MQ and Mom4J
were the two that had support for Python, but Activ MQ seems too
Windows-centric for serious use on Linux. Mom4J looked more promising, but
their implementation was incomplete and their user community was like a ghost
town. There's really no point in investing time in a dead project (unless it
has killer technology). When I discovered mod_pubsub and the REST-based systems
(been on my radar for a long time, but never delved into it) I was happy to
find out that they overcome the confinement limitations of JMS. At this point,
I'm ignorant of the security ramifications that this presents, which I'd like
to find out more. FWIW, Mom4J provided hooks for authentication, but no actual
implementation was present.

I'll probably be looking at some other REST systems, and I'll let you know the
outcome.

Brian


Quoting mod...@li...:

> In the interest of strengthening everyone's knowledge base, what did you 
> compare against and what were your concerns?  I've noticed that 
> xmlBlaster and Activ MQ both offer HTTP oriented methodologies for 
> asynchronous notification.  But I am not sure how robust they are, nor 
> how native/central to the design they are, nor how well they work in 
> javascript contexts.
> 
> A.
> 
> mod...@li... wrote:
> 
> >Adam:
> >
> >  
> >
> >>The question was not bone-headed at all. It's just that sometimes our
> >>Python guy goes into  { while 1: sleep } mode. ;)
> >>    
> >>
> >
> >In that case, I'll continue to poll() the list. Hopefully the results aren't
> too
> >asynchronous!
> >
> >
> >  
> >
> >>Both of those client libraries were experimental prototypes, as is
> >>probably clear from reading the source. Plus, I was kind of learning
> >>Python as we wrote them.
> >>
> >>I'm Cc'ing Ben in case he knows the answer to your embedding question.
> >>    
> >>
> >
> >So is anyone using the Python clients for any interesting applications?
> I've
> >evaluated some other asynchronous messaging technologies, and mod/re-pubsub
> >looks the most promising w.r.t. Python and Internet-wide messaging. It would
> be
> >great if anyone could share some insight into the embedding question, or how
> to
> >harness the non-blocking behavior of the pubsublib.
> >
> >
> >  
> >
> >>I do know that no one is currently working to port the client library
> >>to Twisted; right now all the energy and effort is going into
> >>rewriting the server using Twisted, and hopefully we can learn from
> >>that to do a client rewrite sometime later. If you're feeling
> >>ambitious, feel free to take a crack at doing it with Twisted's
> >>Deferred mechanism...
> >>    
> >>
> >
> >How is the progress on the server side? Is the current version stable enough
> to
> >be making changes that would be compatible with the current server
> development?
> >I have some experience with Twisted, so I wouldn't mind taking a look, but
> I
> >would like to get a sense of the project road map so my work isn't wasted.
> >
> >  
> >
> >>   -- Adam
> >>    
> >>
> >
> >Brian
> >
> >
> >PS Are you Adam Rifkin or another Adam?
> >
> >
> >  
> >
> >>On Apr 5, 2005 4:13 AM,
> >>mod...@li...
> >><mod...@li...> wrote:
> >>    
> >>
> >>>Haven't heard anything back from the list regarding this question. If it
> >>>      
> >>>
> >>is
> >>    
> >>
> >>>because the question was too bone-headed, then there is no remedy.
> >>>      
> >>>
> >>Otherwise,
> >>    
> >>
> >>>the following information might be useful to someone else getting
> started.
> >>>
> >>>There are two python client libraries: pubsublib.py and libkn/libkn.py.
> >>>According to the source, pubsublib.py is a non-blocking implementation,
> >>>      
> >>>
> >>while
> >>    
> >>
> >>>libkn.py is blocking. The examples in apps are built off of pubsublib, so
> >>>      
> >>>
> >>they
> >>    
> >>
> >>>should be non-blocking.
> >>>
> >>>This seems to be true when running as a standalone script, but what about
> >>>      
> >>>
> >>when I
> >>    
> >>
> >>>want to embed? Is the logging package being used to help the debugging
> >>>      
> >>>
> >>process
> >>    
> >>
> >>>or are people using pdb?
> >>>
> >>>Also, I read somewhere that the client is supposed to be using Twisted in
> >>>      
> >>>
> >>the
> >>    
> >>
> >>>near future. Is this still the case, and if so, what's the progress on
> it?
> >>>
> >>>Thanks,
> >>>Brian
> >>>
> >>>
> >>>Quoting mod...@li...:
> >>>
> >>>      
> >>>
> >>>>Hello:
> >>>>
> >>>>I'm trying to use the apps/publish.py module as a starting point for
> >>>>        
> >>>>
> >>writing
> >>    
> >>
> >>>>my
> >>>>own simple publishing lib to work with repubsub. The script runs fine on
> >>>>        
> >>>>
> >>its
> >>    
> >>
> >>>>own, but whenever I try to embed it in other code, I seem to get
> >>>>        
> >>>>
> >>blocking
> >>    
> >>
> >>>>issues and the CPU load spikes.
> >>>>
> >>>>Inside the apps/publish.py script, I looked at the following method
> >>>>        
> >>>>
> >>calls:
> >>    
> >>
> >>>>  asyncore.poll()
> >>>>  scheduler.timeout()
> >>>>  scheduler.run()
> >>>>
> >>>>Now Python's asyncore does not have documentation on the poll function,
> >>>>although
> >>>>in my interpreter, I do see a reference for poll. The two scheduler
> >>>>        
> >>>>
> >>function
> >>    
> >>
> >>>>calls both pass in their implementations. If I comment out that loop, my
> >>>>messages don't get published. So what's going on here?
> >>>>
> >>>>Ideally, I'd like to be able to use Twisted's Deferred mechanism to
> >>>>        
> >>>>
> >>handle
> >>    
> >>
> >>>>the
> >>>>blocking issue, but I'm unsure the best way to approach it without
> >>>>        
> >>>>
> >>heavily
> >>    
> >>
> >>>>modifying the code base.
> >>>>
> >>>>Any help is greatly appreciated.
> >>>>
> >>>>Thanks,
> >>>>Brian
> >>>>
> >>>>--
> >>>>  ====
> >>>>brian lee dae yung
> >>>>biomimetic art and research
> >>>>  mux space . com
> >>>>
> >>>>----------------------------------------------------------------
> >>>>This message was sent using IMP, the Internet Messaging Program.
> >>>>
> >>>>
> >>>>
> >>>>-------------------------------------------------------
> >>>>SF email is sponsored by - The IT Product Guide
> >>>>Read honest & candid reviews on hundreds of IT Products from real users.
> >>>>Discover which products truly live up to the hype. Start reading now.
> >>>>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> >>>>_______________________________________________
> >>>>Mod-pubsub-developer mailing list
> >>>>Mod...@li...
> >>>>https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> >>>>
> >>>>        
> >>>>
> >>>--
> >>>  ====
> >>>brian lee dae yung
> >>>biomimetic art and research
> >>>  mux space . com
> >>>
> >>>----------------------------------------------------------------
> >>>This message was sent using IMP, the Internet Messaging Program.
> >>>
> >>>-------------------------------------------------------
> >>>SF email is sponsored by - The IT Product Guide
> >>>Read honest & candid reviews on hundreds of IT Products from real users.
> >>>Discover which products truly live up to the hype. Start reading now.
> >>>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> >>>_______________________________________________
> >>>Mod-pubsub-developer mailing list
> >>>Mod...@li...
> >>>https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> >>>
> >>>      
> >>>
> >>-------------------------------------------------------
> >>SF email is sponsored by - The IT Product Guide
> >>Read honest & candid reviews on hundreds of IT Products from real users.
> >>Discover which products truly live up to the hype. Start reading now.
> >>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> >>_______________________________________________
> >>Mod-pubsub-developer mailing list
> >>Mod...@li...
> >>https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> >>
> >>    
> >>
> >
> >
> >  
> >
> 
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> 


-- 
  ====
brian lee dae yung
biomimetic art and research
  mux space . com

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

[Mod-pubsub-developer] Mod-pubsub server rewrite.

[<mod...@li...>]

Some people have asked us about what's going on with the mod-pubsub
server rewrite, so here's a few notes. I'm also posting this note on
the blog at

     http://www.mod-pubsub.org/blog/archives/454_Modpubsub_Server_Rewrite__Update.html


Repubsub included a nice improvement in the mod-pubsub python server,
replacing asyncore with the Twisted library. Unfortunately, it still
leaks memory, so we're in the process of raising some cash to pay for
a complete server rewrite.

Here are the components of that project:

1. A complete rewrite of the project source, re-using no original code.
2. Seek maximal use of third-party libraries to minimize the amount of
forked code that needs to be maintained.
3. Removal of deprecated and unused features and code paths.
4. Introduction of an extension framework via a comprehensive set of
registerable hooks.
5. Build a set of reference extensions using the new extension framework.
6. A unit testing suite will be produced.
7. A performance benchmarking tool will produced, designed to measure
(separately) system latency, concurrency and throughput.
8. Full code documentation.

We have so far raised about a third of the money from a few companies
but we still have plenty of room. If you know any companies that would
like to contribute dollars, send them my way.

Our ETA for delivery is the end of June, which should include some
patches to the JavaScript client library as well.

Will post updates when I have them...

Cheers,
   Adam

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

In the interest of strengthening everyone's knowledge base, what did you 
compare against and what were your concerns?  I've noticed that 
xmlBlaster and Activ MQ both offer HTTP oriented methodologies for 
asynchronous notification.  But I am not sure how robust they are, nor 
how native/central to the design they are, nor how well they work in 
javascript contexts.

A.

mod...@li... wrote:

>Adam:
>
>  
>
>>The question was not bone-headed at all. It's just that sometimes our
>>Python guy goes into  { while 1: sleep } mode. ;)
>>    
>>
>
>In that case, I'll continue to poll() the list. Hopefully the results aren't too
>asynchronous!
>
>
>  
>
>>Both of those client libraries were experimental prototypes, as is
>>probably clear from reading the source. Plus, I was kind of learning
>>Python as we wrote them.
>>
>>I'm Cc'ing Ben in case he knows the answer to your embedding question.
>>    
>>
>
>So is anyone using the Python clients for any interesting applications? I've
>evaluated some other asynchronous messaging technologies, and mod/re-pubsub
>looks the most promising w.r.t. Python and Internet-wide messaging. It would be
>great if anyone could share some insight into the embedding question, or how to
>harness the non-blocking behavior of the pubsublib.
>
>
>  
>
>>I do know that no one is currently working to port the client library
>>to Twisted; right now all the energy and effort is going into
>>rewriting the server using Twisted, and hopefully we can learn from
>>that to do a client rewrite sometime later. If you're feeling
>>ambitious, feel free to take a crack at doing it with Twisted's
>>Deferred mechanism...
>>    
>>
>
>How is the progress on the server side? Is the current version stable enough to
>be making changes that would be compatible with the current server development?
>I have some experience with Twisted, so I wouldn't mind taking a look, but I
>would like to get a sense of the project road map so my work isn't wasted.
>
>  
>
>>   -- Adam
>>    
>>
>
>Brian
>
>
>PS Are you Adam Rifkin or another Adam?
>
>
>  
>
>>On Apr 5, 2005 4:13 AM,
>>mod...@li...
>><mod...@li...> wrote:
>>    
>>
>>>Haven't heard anything back from the list regarding this question. If it
>>>      
>>>
>>is
>>    
>>
>>>because the question was too bone-headed, then there is no remedy.
>>>      
>>>
>>Otherwise,
>>    
>>
>>>the following information might be useful to someone else getting started.
>>>
>>>There are two python client libraries: pubsublib.py and libkn/libkn.py.
>>>According to the source, pubsublib.py is a non-blocking implementation,
>>>      
>>>
>>while
>>    
>>
>>>libkn.py is blocking. The examples in apps are built off of pubsublib, so
>>>      
>>>
>>they
>>    
>>
>>>should be non-blocking.
>>>
>>>This seems to be true when running as a standalone script, but what about
>>>      
>>>
>>when I
>>    
>>
>>>want to embed? Is the logging package being used to help the debugging
>>>      
>>>
>>process
>>    
>>
>>>or are people using pdb?
>>>
>>>Also, I read somewhere that the client is supposed to be using Twisted in
>>>      
>>>
>>the
>>    
>>
>>>near future. Is this still the case, and if so, what's the progress on it?
>>>
>>>Thanks,
>>>Brian
>>>
>>>
>>>Quoting mod...@li...:
>>>
>>>      
>>>
>>>>Hello:
>>>>
>>>>I'm trying to use the apps/publish.py module as a starting point for
>>>>        
>>>>
>>writing
>>    
>>
>>>>my
>>>>own simple publishing lib to work with repubsub. The script runs fine on
>>>>        
>>>>
>>its
>>    
>>
>>>>own, but whenever I try to embed it in other code, I seem to get
>>>>        
>>>>
>>blocking
>>    
>>
>>>>issues and the CPU load spikes.
>>>>
>>>>Inside the apps/publish.py script, I looked at the following method
>>>>        
>>>>
>>calls:
>>    
>>
>>>>  asyncore.poll()
>>>>  scheduler.timeout()
>>>>  scheduler.run()
>>>>
>>>>Now Python's asyncore does not have documentation on the poll function,
>>>>although
>>>>in my interpreter, I do see a reference for poll. The two scheduler
>>>>        
>>>>
>>function
>>    
>>
>>>>calls both pass in their implementations. If I comment out that loop, my
>>>>messages don't get published. So what's going on here?
>>>>
>>>>Ideally, I'd like to be able to use Twisted's Deferred mechanism to
>>>>        
>>>>
>>handle
>>    
>>
>>>>the
>>>>blocking issue, but I'm unsure the best way to approach it without
>>>>        
>>>>
>>heavily
>>    
>>
>>>>modifying the code base.
>>>>
>>>>Any help is greatly appreciated.
>>>>
>>>>Thanks,
>>>>Brian
>>>>
>>>>--
>>>>  ====
>>>>brian lee dae yung
>>>>biomimetic art and research
>>>>  mux space . com
>>>>
>>>>----------------------------------------------------------------
>>>>This message was sent using IMP, the Internet Messaging Program.
>>>>
>>>>
>>>>
>>>>-------------------------------------------------------
>>>>SF email is sponsored by - The IT Product Guide
>>>>Read honest & candid reviews on hundreds of IT Products from real users.
>>>>Discover which products truly live up to the hype. Start reading now.
>>>>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
>>>>_______________________________________________
>>>>Mod-pubsub-developer mailing list
>>>>Mod...@li...
>>>>https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
>>>>
>>>>        
>>>>
>>>--
>>>  ====
>>>brian lee dae yung
>>>biomimetic art and research
>>>  mux space . com
>>>
>>>----------------------------------------------------------------
>>>This message was sent using IMP, the Internet Messaging Program.
>>>
>>>-------------------------------------------------------
>>>SF email is sponsored by - The IT Product Guide
>>>Read honest & candid reviews on hundreds of IT Products from real users.
>>>Discover which products truly live up to the hype. Start reading now.
>>>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
>>>_______________________________________________
>>>Mod-pubsub-developer mailing list
>>>Mod...@li...
>>>https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
>>>
>>>      
>>>
>>-------------------------------------------------------
>>SF email is sponsored by - The IT Product Guide
>>Read honest & candid reviews on hundreds of IT Products from real users.
>>Discover which products truly live up to the hype. Start reading now.
>>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
>>_______________________________________________
>>Mod-pubsub-developer mailing list
>>Mod...@li...
>>https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
>>
>>    
>>
>
>
>  
>

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Adam:

> The question was not bone-headed at all. It's just that sometimes our
> Python guy goes into  { while 1: sleep } mode. ;)

In that case, I'll continue to poll() the list. Hopefully the results aren't too
asynchronous!


> Both of those client libraries were experimental prototypes, as is
> probably clear from reading the source. Plus, I was kind of learning
> Python as we wrote them.
> 
> I'm Cc'ing Ben in case he knows the answer to your embedding question.

So is anyone using the Python clients for any interesting applications? I've
evaluated some other asynchronous messaging technologies, and mod/re-pubsub
looks the most promising w.r.t. Python and Internet-wide messaging. It would be
great if anyone could share some insight into the embedding question, or how to
harness the non-blocking behavior of the pubsublib.


> I do know that no one is currently working to port the client library
> to Twisted; right now all the energy and effort is going into
> rewriting the server using Twisted, and hopefully we can learn from
> that to do a client rewrite sometime later. If you're feeling
> ambitious, feel free to take a crack at doing it with Twisted's
> Deferred mechanism...

How is the progress on the server side? Is the current version stable enough to
be making changes that would be compatible with the current server development?
I have some experience with Twisted, so I wouldn't mind taking a look, but I
would like to get a sense of the project road map so my work isn't wasted.

>    -- Adam

Brian


PS Are you Adam Rifkin or another Adam?


> On Apr 5, 2005 4:13 AM,
> mod...@li...
> <mod...@li...> wrote:
> > Haven't heard anything back from the list regarding this question. If it
> is
> > because the question was too bone-headed, then there is no remedy.
> Otherwise,
> > the following information might be useful to someone else getting started.
> > 
> > There are two python client libraries: pubsublib.py and libkn/libkn.py.
> > According to the source, pubsublib.py is a non-blocking implementation,
> while
> > libkn.py is blocking. The examples in apps are built off of pubsublib, so
> they
> > should be non-blocking.
> > 
> > This seems to be true when running as a standalone script, but what about
> when I
> > want to embed? Is the logging package being used to help the debugging
> process
> > or are people using pdb?
> > 
> > Also, I read somewhere that the client is supposed to be using Twisted in
> the
> > near future. Is this still the case, and if so, what's the progress on it?
> > 
> > Thanks,
> > Brian
> > 
> > 
> > Quoting mod...@li...:
> > 
> > > Hello:
> > >
> > > I'm trying to use the apps/publish.py module as a starting point for
> writing
> > > my
> > > own simple publishing lib to work with repubsub. The script runs fine on
> its
> > > own, but whenever I try to embed it in other code, I seem to get
> blocking
> > > issues and the CPU load spikes.
> > >
> > > Inside the apps/publish.py script, I looked at the following method
> calls:
> > >
> > >   asyncore.poll()
> > >   scheduler.timeout()
> > >   scheduler.run()
> > >
> > > Now Python's asyncore does not have documentation on the poll function,
> > > although
> > > in my interpreter, I do see a reference for poll. The two scheduler
> function
> > > calls both pass in their implementations. If I comment out that loop, my
> > > messages don't get published. So what's going on here?
> > >
> > > Ideally, I'd like to be able to use Twisted's Deferred mechanism to
> handle
> > > the
> > > blocking issue, but I'm unsure the best way to approach it without
> heavily
> > > modifying the code base.
> > >
> > > Any help is greatly appreciated.
> > >
> > > Thanks,
> > > Brian
> > >
> > > --
> > >   ====
> > > brian lee dae yung
> > > biomimetic art and research
> > >   mux space . com
> > >
> > > ----------------------------------------------------------------
> > > This message was sent using IMP, the Internet Messaging Program.
> > >
> > >
> > >
> > > -------------------------------------------------------
> > > SF email is sponsored by - The IT Product Guide
> > > Read honest & candid reviews on hundreds of IT Products from real users.
> > > Discover which products truly live up to the hype. Start reading now.
> > > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> > > _______________________________________________
> > > Mod-pubsub-developer mailing list
> > > Mod...@li...
> > > https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> > >
> > 
> > --
> >   ====
> > brian lee dae yung
> > biomimetic art and research
> >   mux space . com
> > 
> > ----------------------------------------------------------------
> > This message was sent using IMP, the Internet Messaging Program.
> > 
> > -------------------------------------------------------
> > SF email is sponsored by - The IT Product Guide
> > Read honest & candid reviews on hundreds of IT Products from real users.
> > Discover which products truly live up to the hype. Start reading now.
> > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> > _______________________________________________
> > Mod-pubsub-developer mailing list
> > Mod...@li...
> > https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> >
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> 


-- 
  ====
brian lee dae yung
biomimetic art and research
  mux space . com

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Hi Brian,

The question was not bone-headed at all. It's just that sometimes our
Python guy goes into  { while 1: sleep } mode. ;)

Both of those client libraries were experimental prototypes, as is
probably clear from reading the source. Plus, I was kind of learning
Python as we wrote them.

I'm Cc'ing Ben in case he knows the answer to your embedding question.

I do know that no one is currently working to port the client library
to Twisted; right now all the energy and effort is going into
rewriting the server using Twisted, and hopefully we can learn from
that to do a client rewrite sometime later. If you're feeling
ambitious, feel free to take a crack at doing it with Twisted's
Deferred mechanism...

   -- Adam



On Apr 5, 2005 4:13 AM,
mod...@li...
<mod...@li...> wrote:
> Haven't heard anything back from the list regarding this question. If it is
> because the question was too bone-headed, then there is no remedy. Otherwise,
> the following information might be useful to someone else getting started.
> 
> There are two python client libraries: pubsublib.py and libkn/libkn.py.
> According to the source, pubsublib.py is a non-blocking implementation, while
> libkn.py is blocking. The examples in apps are built off of pubsublib, so they
> should be non-blocking.
> 
> This seems to be true when running as a standalone script, but what about when I
> want to embed? Is the logging package being used to help the debugging process
> or are people using pdb?
> 
> Also, I read somewhere that the client is supposed to be using Twisted in the
> near future. Is this still the case, and if so, what's the progress on it?
> 
> Thanks,
> Brian
> 
> 
> Quoting mod...@li...:
> 
> > Hello:
> >
> > I'm trying to use the apps/publish.py module as a starting point for writing
> > my
> > own simple publishing lib to work with repubsub. The script runs fine on its
> > own, but whenever I try to embed it in other code, I seem to get blocking
> > issues and the CPU load spikes.
> >
> > Inside the apps/publish.py script, I looked at the following method calls:
> >
> >   asyncore.poll()
> >   scheduler.timeout()
> >   scheduler.run()
> >
> > Now Python's asyncore does not have documentation on the poll function,
> > although
> > in my interpreter, I do see a reference for poll. The two scheduler function
> > calls both pass in their implementations. If I comment out that loop, my
> > messages don't get published. So what's going on here?
> >
> > Ideally, I'd like to be able to use Twisted's Deferred mechanism to handle
> > the
> > blocking issue, but I'm unsure the best way to approach it without heavily
> > modifying the code base.
> >
> > Any help is greatly appreciated.
> >
> > Thanks,
> > Brian
> >
> > --
> >   ====
> > brian lee dae yung
> > biomimetic art and research
> >   mux space . com
> >
> > ----------------------------------------------------------------
> > This message was sent using IMP, the Internet Messaging Program.
> >
> >
> >
> > -------------------------------------------------------
> > SF email is sponsored by - The IT Product Guide
> > Read honest & candid reviews on hundreds of IT Products from real users.
> > Discover which products truly live up to the hype. Start reading now.
> > http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> > _______________________________________________
> > Mod-pubsub-developer mailing list
> > Mod...@li...
> > https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> >
> 
> --
>   ====
> brian lee dae yung
> biomimetic art and research
>   mux space . com
> 
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
>

Re: [Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Haven't heard anything back from the list regarding this question. If it is
because the question was too bone-headed, then there is no remedy. Otherwise,
the following information might be useful to someone else getting started.

There are two python client libraries: pubsublib.py and libkn/libkn.py.
According to the source, pubsublib.py is a non-blocking implementation, while
libkn.py is blocking. The examples in apps are built off of pubsublib, so they
should be non-blocking.

This seems to be true when running as a standalone script, but what about when I
want to embed? Is the logging package being used to help the debugging process
or are people using pdb?

Also, I read somewhere that the client is supposed to be using Twisted in the
near future. Is this still the case, and if so, what's the progress on it?

Thanks,
Brian


Quoting mod...@li...:

> Hello:
> 
> I'm trying to use the apps/publish.py module as a starting point for writing
> my
> own simple publishing lib to work with repubsub. The script runs fine on its
> own, but whenever I try to embed it in other code, I seem to get blocking
> issues and the CPU load spikes.
> 
> Inside the apps/publish.py script, I looked at the following method calls:
> 
>   asyncore.poll()
>   scheduler.timeout()
>   scheduler.run()
> 
> Now Python's asyncore does not have documentation on the poll function,
> although
> in my interpreter, I do see a reference for poll. The two scheduler function
> calls both pass in their implementations. If I comment out that loop, my
> messages don't get published. So what's going on here? 
> 
> Ideally, I'd like to be able to use Twisted's Deferred mechanism to handle
> the
> blocking issue, but I'm unsure the best way to approach it without heavily
> modifying the code base.
> 
> Any help is greatly appreciated.
> 
> Thanks,
> Brian
> 
> -- 
>   ====
> brian lee dae yung
> biomimetic art and research
>   mux space . com
> 
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
> 
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Mod-pubsub-developer mailing list
> Mod...@li...
> https://lists.sourceforge.net/lists/listinfo/mod-pubsub-developer
> 


-- 
  ====
brian lee dae yung
biomimetic art and research
  mux space . com

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Re: [Mod-pubsub-developer] Replaying events?

[<mod...@li...>]

So not to be unhelpful, but this is the kind of thing we're trying to move to
the new hook mechanism in the server rewrite.  Then it will be your problem,
not ours.  :-)  JP


--- mod...@li... wrote:
> Hi,
> 
> I asked this question before but didn't get an answer. How do I get the 
> Repubsub server to replay events in a certain topic to the javascript 
> client. There was a way to do this in Old Mod pub sub but not in the new 
> javascript client library

[Mod-pubsub-developer] Replaying events?

[<mod...@li...>]

Hi,

I asked this question before but didn't get an answer. How do I get the 
Repubsub server to replay events in a certain topic to the javascript 
client. There was a way to do this in Old Mod pub sub but not in the new 
javascript client library

Thanks

Steven

[Mod-pubsub-developer] .Net connector Assembly failing to load without Visual Studio

[<mod...@li...>]

Hi ,
	I am having a problem running an application that uses the .Net
connector on a machine that has only the .Net framework 1.1 installed on it.
I get a FileNotFoundException That the file LibKNDotNet_71 or one of its
dependcies was not found. (Even though the dll is in the same directory as
the exe)
	Earlier, the Assembly version number in the DotNet_71 project was
1.0.* . But the output produced is always 1.0.0.1. I tried redirecting the
application to 1.0.0.1 but that threw the same exception. Next, I changed
the assembly version number to 1.0.0.1, and recompiled the connector project
and used that reference but the problem still persisted. 
	The application runs without any problems if I have Visual Studio
installed. I have checked the dependecies and it only shows mscorlib and
LIbKNDotnet_71
	What could be the problem? Any help would be appreciated.

Regards,
Rohit Khetan.



This message is free from Virus - IMSS

RE: [Mod-pubsub-developer] Re: Quick question...

[<mod...@li...>]

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dus-ascii">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
6.0.4417.0">
<TITLE>RE: [Mod-pubsub-developer] Re: Quick question...</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<BR>

<P><FONT SIZE=3D2>&gt; There is an sourceforge project called 'destiny' =
that is an HTTP </FONT>

<BR><FONT SIZE=3D2>&gt; bridge to JMS.</FONT>
</P>

<P><FONT SIZE=3D2>Excellent suggestion.</FONT>
</P>

<P><FONT SIZE=3D2>&gt; You could use that as a starting point to =
implement that same HTTP API </FONT>

<BR><FONT SIZE=3D2>&gt; used by mod-pubsub. </FONT>
</P>

<P><FONT SIZE=3D2>Were can I find (either files in distribution or docs) =
details on the HTTP API?</FONT>

<BR><FONT SIZE=3D2>Is there any info on the mechanism used by mod-pubsub =
for the keep alive connection?</FONT>

<BR><FONT SIZE=3D2>I am assume this is accomplished by both, having =
'keep-alive' plus other info in the header?</FONT>
</P>

<P><FONT SIZE=3D2>Oliver Muoto</FONT>

<BR><FONT SIZE=3D2>ol...@mu...</FONT>
</P>

</BODY>
</HTML>

Re: [Mod-pubsub-developer] Re: Quick question...

[<mod...@li...>]

Uh... this email wasn't signed, but I think it's Dierken?  If so, hi Mike!

Just to clarify, the Mod-pubsub SERVER API has not changed at all yet, except
for one tiny thing that was necessary for the Flash client but should not
impact anything else.  The JavaScript API is the thing that changed
significantly.  JP


--- mod...@li... wrote:

> There is an sourceforge project called 'destiny' that is an HTTP bridge to
> JMS.
> You could use that as a starting point to implement that same HTTP API used
> by mod-pubsub. There's nothing magic about the server implementation of
> mod-pubsub, it's all in the protocol. If mod-pubsub had a multi-server
> protocol for a constellation of messaging servers, that would be more work
> to 'port', but still doable.
> 
> The java client I built is on the old mod-pubsub API, I don't think it's
> compatible with the new server implementation.
> There are actually two java client libraries in CVS for mod-pubsub. One has
> lots of unit tests. Mine doesn't... but still works.
> 
> Let me know if you need advice or additional work for a java client or java
> server - i'd love to help implement a java server.
> 
> I'd also like to implement an RSS format for 'change-lists' (lists of recent
> events) that use 'next' and 'prev' links to traverse past and future
> history - I think that's the correct approach for a REST based event
> notification system.

[Mod-pubsub-developer] Re: Quick question...

[<mod...@li...>]

There is an sourceforge project called 'destiny' that is an HTTP bridge to
JMS.
You could use that as a starting point to implement that same HTTP API used
by mod-pubsub. There's nothing magic about the server implementation of
mod-pubsub, it's all in the protocol. If mod-pubsub had a multi-server
protocol for a constellation of messaging servers, that would be more work
to 'port', but still doable.

The java client I built is on the old mod-pubsub API, I don't think it's
compatible with the new server implementation.
There are actually two java client libraries in CVS for mod-pubsub. One has
lots of unit tests. Mine doesn't... but still works.

Let me know if you need advice or additional work for a java client or java
server - i'd love to help implement a java server.

I'd also like to implement an RSS format for 'change-lists' (lists of recent
events) that use 'next' and 'prev' links to traverse past and future
history - I think that's the correct approach for a REST based event
notification system.


----- Original Message ----- 
From: "Adam Rifkin" <ifi...@gm...>
To: "Oliver Muoto" <ol...@mu...>
Cc: <mod...@li...>; "Kragen Sitaker"
<ksi...@co...>; <mdi...@ho...>; "Gregory Burd"
<gre...@gm...>; "Benjamin C. Wiley Sittler" <bsi...@gm...>
Sent: Friday, April 01, 2005 10:16 AM
Subject: Re: Quick question...


> Hi Oliver,
>
> Thanks for your comments.
>
> I believe experimental client libraries for Java already exist in
> mod-pubsub. Mike Dierken, who is Cc'd, might know more.
>
> As far as the server being ported to Java, no one has yet attempted.
> One quick thing to try is using Jython to compile the Python to Java
> bytecodes, though I don't believe anyone has tried this yet.
>
> I'm Cc'ing the mod-pubsub-developer list in case anyone else has heard
> about any attempts to port to Java...
>
> Cheers,
>   Adam
>
> On Apr 1, 2005 8:26 AM, Oliver Muoto <ol...@mu...> wrote:
> >
> >
> > Adam
> > Need your advice. I want (need) to convert Mod-pubsub to Java! Doing a
> > decent amount of AJAX development and have developed a need for
real-time
> > data boss. Have no problem using Mod-pubsub but some clients has a major
> > problem with "lesser languages" ;) .. No offence, but given most of
their
> > other infrastructure is J2EE, they are not willing to mix 'platforms'.
Any
> > suggestions? I have tried to do it 'on the cheap' but none of my
outsourced
> > development teams seem to get it, everyone things I am trying to build a
JMS
> > implementation ;(. Before I have to shell out real money, wanted to
check in
> > with you. Any plans to port to Java? If you do, I think you open a WHOLE
new
> > (and large) market (if you can call it that) for the solution. Most
large
> > systems today are still J2EE, and in that world there is no mix and
match,
> > there is but Java.
> >
> > Oliver
> >
> > Oliver Muoto
> > Co-founder, Epicentric
> > e: ol...@mu...
> > c: 650-787-5567
> > f: 603-258-8950
> > http://www.muoto.com?s=o
> >
> > Metablocks
> > Advanced UI and App Design
> > http://www.metablocks.com/jump?s=o
> >
> > MerchantCircle
> > Want free advertising for your business
> > http://www.merchantcircle.com/free
> >
> > Vignette Completes Acquisition of Epicentric:
> > http://www.vignette.com/epicentric
>

[Mod-pubsub-developer] Non-blocking publishing in Python

[<mod...@li...>]

Hello:

I'm trying to use the apps/publish.py module as a starting point for writing my
own simple publishing lib to work with repubsub. The script runs fine on its
own, but whenever I try to embed it in other code, I seem to get blocking
issues and the CPU load spikes.

Inside the apps/publish.py script, I looked at the following method calls:

  asyncore.poll()
  scheduler.timeout()
  scheduler.run()

Now Python's asyncore does not have documentation on the poll function, although
in my interpreter, I do see a reference for poll. The two scheduler function
calls both pass in their implementations. If I comment out that loop, my
messages don't get published. So what's going on here? 

Ideally, I'd like to be able to use Twisted's Deferred mechanism to handle the
blocking issue, but I'm unsure the best way to approach it without heavily
modifying the code base.

Any help is greatly appreciated.

Thanks,
Brian

-- 
  ====
brian lee dae yung
biomimetic art and research
  mux space . com

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

Re: [Mod-pubsub-developer] Publishing with the Python client

[<mod...@li...>]

Adam:

> > Nevermind. I looked at the publish.py and subscribe.py to see what's going
> on.
> > The semantics are a bit confusing because the payload size is being used
> to
> > generate the payload. Perhaps an echo pub and sub would have been easier
> to
> > understand (I actually did just that with the example).
> 
> You can actually blame the unclear thinking on me. I really was
> learning Python as I was writing those toy apps -- and it shows. ;)

No worries, we all start somewhere.  It might be good to have some updated
examples or cookbook scripts to get people started. For example, one thing I'd
like to know right now is how to suppress all of the status messages when using
the lib. I'm actually using the tests/rss_scroller.html as a base template and
playing around with things in there and it would be great to have it running
without all of the status and log messages. Any suggestions?

Also, is the best way to load the libraries with the /kn?do_method=lib call?
Which directory does the server serve the js files from? I'm unclear about the
difference between the kn_client* directories.


> > As for why it wasn't updating on the JS client, I'm not sure. After some
> > fiddling though, it's working now.
> 
> If you are ever able to isolate the behavior, we'd love to know.

It was probably user error. Perhaps I didn't have the JS client subscribed to
the right topic. 


> If you want some older, more experimental (read: *very* prototypey)
> apps to play with, feel free to download the April 20, 2004 tarball
> from Sourceforge (v0.994 -- long before the much cooler and much more
> bug free repubsub came along):

LOL. I would actually prefer some newer, less experimental apps to play with.
The current directory structure is a little confusing, particularly since it
acts as the web directory. It might make sense to encapsulate the web dir and
explicitly create folders for the examples and client libs. An installer might
be nice too, so the server can sit somewher within /opt or /var. I can help out
in this area if anyone is interested. 

Regards,
Brian

-- 
  ====
brian lee dae yung
biomimetic art and research
  mux space . com

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.