Daniel Henninger | 1 Feb 2007 17:30
Picon

Re: Re: (no subject)

Aww crap, I totally lost this message.  =)

On Jan 29, 2007, at 10:46 AM, Alexander V. Butenko wrote:

> Mon, 29 Jan 2007 you wrote:
>
>>> 3. A patch for correct detecting of russian text in statuses is  
>>> added
>>> (from a bugtrack. tested for a half a year)
>> It isn't the same patch from the bugtracker is it?  (that was  
>> submitted years ago)  There was a patch submitted that, while it  
>> fixed up
>> russian nicely, it screwed over all the other languages.  ;D  (but  
>> that was more than just statuses)
>
> Yes, it is. So u might not to put to svn.

ok  =) thanks!  (I'll check and make sure it's exactly the one I was  
thinking about)

>
>>> With this patch pyicq was runned for weeks without a memory leaks.
>>> The problem of leaks is in web panel code and AvatarCache Factory.
>> I'm confused, does this disable the web interface altogether?
>
> I was making this patch in a hurry, so first i begin just to delete  
> web
> stuff, instead of adding ifs, so as u see web interface functional is
> just ugly ripped instead of adding ifs. I was going to fix a patch  
> when
(Continue reading)

Chris Carlin | 6 Feb 2007 22:13
Favicon

Gentoo ebuilds

I don't know how many people this applies to, but I've (finally) updated 
the Gentoo ebuilds of pyaim-t and pyicq-t to version 0.8. A few people 
have been asking for these packages for a couple of months. Grab the 
gentoo-sunrise overlay and have fun.

Sorry for the delay; in my defense phone companies suck.

~Chris
Michael Dyrna | 12 Feb 2007 11:30
Picon

Experience with transports on large sites?

Hi everyone,

Who has deployed a py-transports installation with a very high number of 
users and can share their experience with me? Say several thousand users 
registered und several hundert users logged in concurrently?

I am especially wondering about clustering. Does anyone have an idea 
other than

* using XCP's clustering feature (since we are using ejabberd here)

* deploying more than one transport instance, each with its own JID 
(allows a simple kind of load balancing but not redundancy since they 
would all have their own user base).

Thanks for your advice,

-Michael
Elder Sales | 12 Feb 2007 13:34
Picon

Py Transports Corporate use

Hi folks.

 Iam brazilian and Iam having experience to using PyMSNt fine but now i wish
to use it as a corporate mode with about 150 users at same time.
Can i reach this number of user in my company ?
 How the most number of users have all you using Py Transports?
 Ps: Iam working with Jabberd2 and Mysql 5

 Best regards for all

--

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.441 / Virus Database: 268.17.32/677 - Release Date: 8/2/2007
21:04
Daniel Henninger | 12 Feb 2007 21:13
Picon

PyAIMt Welcome Chris Carlin!

Hi folk!

Chris Carlin has agreed to help out with PyAIMt time permitting so  
I'm proud to say we now have more than just me on the project!  =D   
Yay Chris!  Chris has submitted some great code in the past that  
PyAIMt has been using for quite some time now.  (like the OSCAR rate  
limiting funcctionality for example)  Anyway, I wanted to let you all  
know so that you don't wonder why he may start discussing things with  
you all "like a developer".  ;)

Daniel
Pedro Melo | 13 Feb 2007 00:51
Picon

Re: Experience with transports on large sites?

Hi,

On Feb 12, 2007, at 10:30 AM, Michael Dyrna wrote:

> Who has deployed a py-transports installation with a very high  
> number of
> users and can share their experience with me? Say several thousand  
> users
> registered und several hundert users logged in concurrently?
>
> I am especially wondering about clustering. Does anyone have an idea
> other than
>
> * using XCP's clustering feature (since we are using ejabberd here)

A little birdie told me that ejabberd is getting domain_balancing ;)  
that does exactly what you want for MSN for ex.

Don't know when they will roll that into a release, but it's on SVN.  
Pull the guide from SVN and search for domain_balancing.

> * deploying more than one transport instance, each with its own JID
> (allows a simple kind of load balancing but not redundancy since they
> would all have their own user base).

I would not recommend this: it's ugly and not really needed.

But what do you want to know exactly? I see around 300k registered  
MSN users, and more than 5k online at any point in time. the new 0.11  
release of pymsnt is much better than the 0.10. If you see memory  
(Continue reading)

Alexander Sulfrian | 13 Feb 2007 14:43

MySQL server has gone away

Hi,
I use pyICQ-t for some days now. Because I only have a few users, that
sleeps all at night and turn off their PC, I noticed, that after time
with no activity on the transport, the MySQL server closes the
connection. The pyICQ-t tries to recover the connection with
self.db.ping() but this don't work.
One way is to change the timeout of the MySQL server to something very
big, that never get reached. But I think thats only a workaround. I
tried already to modify the source in that way, that the transport
will reconnect if the connection is closed. But because I don't now
much about python, I don't get it. I tried to use the result of ping()
to check weather the connection is up or down. But even on ping() the
transport get an exception. Can you please give me an example how to
catch an exception in python? In c++ I would use something like try and
catch. But I don't now the right syntax in python.

Thanks
Alexander Sulfrian

attached:
logfile after the MySQL server closed the connection (this messages are
repeated every 2 seconds)
[2007-02-13 14:14:23] Starting factory <twisted.xish.xmlstream.XmlStreamFactory instance at 0xb7d8fe8c>
[2007-02-13 14:14:23] INFO ::  :: PyTransport :: componentConnected :: {'xmlstream':
<twisted.xish.xmlstream.XmlStream instance at 0xb673ce8c>} :: 
[2007-02-13 14:14:23] Traceback (most recent call last):
          File "/usr/lib/python2.4/site-packages/twisted/python/log.py", line 53, in callWithLogger
            return callWithContext({"system": lp}, func, *args, **kw)
          File "/usr/lib/python2.4/site-packages/twisted/python/log.py", line 38, in callWithContext
(Continue reading)

Elder Sales | 13 Feb 2007 16:43
Picon

Using Avatars

Hi folk!

 Iam trying PyMSNt in my work, its runing with jabberd2 and Mysql on
Slackware 10.2, but now users is asking me for if there is a way to use
avatars just like MSN Clients.
 I have searching in the web without sucess.
 Can anyone tell me if there is a way?

 Iam beginer on linux world and need some help.

 Best regards for all.

--

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.441 / Virus Database: 268.17.37/682 - Release Date: 12/2/2007
13:23
Pedro Melo | 13 Feb 2007 21:27
Picon

Re: Using Avatars

Hi,

On Feb 13, 2007, at 3:43 PM, Elder Sales wrote:

> Hi folk!
>
>  Iam trying PyMSNt in my work, its runing with jabberd2 and Mysql on
> Slackware 10.2, but now users is asking me for if there is a way to  
> use
> avatars just like MSN Clients.

you jabber users?

Sure, they need a client that supports JEP-0153. Don't have a list of- 
hand but iChat does, I think Spark also does it. Gaim-based ones  
almost work, I've had issues with them in the past, don't know what  
happening now. Gajim should work also.

Probably more, don't know.

Best regards,
--
HIId: Pedro Melo
SMTP: melo <at> co.sapo.pt
XMPP: pedro.melo <at> sapo.pt
Scott Dial | 15 Feb 2007 09:23
Favicon

PyAIM-t scheduler creates too many threads

Greetings,

I currently run a Twisted application that provides AIM connectivity. A 
while back I caught that you folks were maintaining the OSCAR protocol 
independently (which is awesome, thank you). I had been maintaining my 
own local version, as well. The main differences are ones focused on 
stability. I've added a variety try/excepts that are errors I've 
discovered via extensive use and logging. Perhaps your project would be 
interested in that aspect of my code.

My real intent in email is to frown on the 0.8 release which 
incorporates a scheduler. This scheduler uses several independent 
threads to defer tasks, but we are using Twisted. Why are you not using 
a callLater approach instead of time.sleep!? In my application, the 
number of threads grows too large and I am unable to create new threads 
for everyone due to some limit (I've not investigated this limit). I 
fear that I will have to maintain a local version.

I've rewrote the "QueueThread" class to not actually be a thread, but 
rather a callLater-based queue. If you are interested in that, I can 
provide this code to you.

-Scott

--

-- 
Scott Dial
scott <at> scottdial.com
scodial <at> cs.indiana.edu

Gmane