Bug 405632

Created attachment 225490 [details]
Screenshot

Created attachment 225491 [details]
fix

This is a patch I found somewhere on the web.  I've tested it and can confirm that it helps.

the patch is simple, ok to create an online update?

Upstream ticket tracking this issue:

* http://developer.pidgin.im/ticket/6220

There are other patches (which are even less intrusive), couldn't find the SVN/GIT (whatever they're working on) to see what upstream decided to use for the already released bug fix version 2.4.3.

After diffing the fixed upstream version of oscar.h with the (bad) version pidgin-2.4.1 (OS11.0) it looks that they took the patch I already attached in this bug.

*** Bug 405588 has been marked as a duplicate of this bug. ***

Just started to work on backporting.

Stanislav, if I didn't miss anything the patch applied cleanly (probably some fuzz, but that's it) on top of our version which ships with OS11.0?

Yes, it applies cleanly on top of 11.0 and it is exactly equal to the 2.4.2->2.4.3 diff. In 10.3 and older, it will be more complicated, as the change may potentially break AIM. Is there anybody with AIM account?

Internal server is down, so I cannot test your build just now.

Actually, CLIENTINFO_ICQ6_6_0_6059 is not used anywhere in the code. Removing it from the patch.

I am starting to put packages into http://pack.suse.cz/sbrabec/bug405632/pidgin/

Please test and report. Update of libpurple is sufficient.

11.0: Tested with ICQ, AIM should not be affected but test would be welcome

Note: When I logged to ICQ first time after installation of patched libpurple, I got the error message once again. But after pression Relogin, everything works OK.

Anja stand-in speaking: yes. SWAMPID: 18438, 
please submit fixed pavkages and patchinfo(s)

(In reply to comment #11 from Stanislav Brabec)

> Actually, CLIENTINFO_ICQ6_6_0_6059 is not used anywhere in the code. Removing
> it from the patch.

Right, but this applies for the other CLIENTINFO_* defines, too.  Only CLIENTINFO_PURPLE_AIM and CLIENTINFO_PURPLE_ICQ are being used, the rest ($ grep -r CLIENTINFO_ * |wc -l results in 22) are defined but not used.

I have just created AIM account.

11.0 ICQ: login works
11.0 AIM: login works

10.3-shortpatch ICQ: login works
10.3-shortpatch AIM: login works

Note: 10.3-shortpatch changes string sent to AIM. Older versions (gaim) and newer pidgin versions differentiate client itentification string AIM and ICQ. Should I backport this code to keep AIM behavior exactly same as before?

pidgin submitted to:
11.0
10.3 (two alternatives)
SLE10-SP2
(Factory will follow with a version update.)

gaim submitted to:
10.2
SLE10-SP2
SLES10
SLES9-SLD
(SLES8-SLED was not done)

pidgin:

I have created another patch for 10.3 - it sends exactly the same string to AIM as before and changes only ICQ string.

10.3 ICQ: login works
10.3 AIM: login works

SLE10-SP2: Uses the same patch as 11.0.


GAIM:

http://pack.suse.cz/sbrabec/bug405632/gaim/

10.2 ICQ: login works
10.2 AIM: login works

All older versions: code and patch are the same. Unless the protocol is really incompatible with NLD9, it should work.

When does the patch is supposed to appear in the update repository?

Regards,
Alberto

After over a month of testing the patch, it still has not been committed to the update repo, hence rendering Pidgin on openSUSE useless for ICQ. Maybe somebody just forgot about it?

Ping! What´s the status? Any reasons why the updates aren´t in the update repository yet?