Client-side buddy list support missing for certain libpurple protocols that require it

[Michał Siejak]

Currently bitlbee assumes that all libpurple protocols support server-side buddy list that is automatically downloaded from the server upon connection. That is not true for certain protocols like "gg" (Gadu-Gadu). This makes using those protocols with bitlbee impractical since one would need to readd all buddies by hand after every bitlbee restart.

I've written a patch for bitlbee which fixes this issue by adding client-side buddy list support for those protocols that require it (currently "gg" only). I've been using this patched version of bitlbee for quite some time now and everything seems to be working fine and stable. Please review and (hopefully) merge with the source tree.

[wilmer]

Ok, finally looking at this now. Looks good overall, but I think there's a setting that enables server-side contacts for gg in which case your patch should *not* be activated, right?

Also, nit: When the user adds a contact to his/her list, by default a nick entry isn't saved unless the user's manually renamed to something else than the default nick generated by BitlBee, either using the rename command or using an additional argument to add. Maybe your patch should change that behaviour for local-contacts accounts?

[AndrzejL]

Unfortunately this patch stopped working when upgrading from version 3.2 to version 3.2.1. I have asked for help on the bitlbee irc channel and dx was trying to rewrite the patch: ​http://ix.io/9RQ but even tho the bitlbee compiles just fine when using it the contacts for gadu-gadu network are not being loaded... I am hoping someone can continue working on the patch from dx till it works again. Maybe even this code could be later (when tested / working) added to the next stable release. It's extremely important for some users.

Thanks in advance.

AndrzejL

[AndrzejL]

Errr... nevermind... The patch created by dx works like a charm... I forgot to do few very simple things like...

chown -R bitlbee:bitlbee /var/lib/bitlbee

after re-compiling / installation. I guess wrong permissions were stopping Bitlbee from properly reading the contacts list for gadu-gadu...

also

mkdir /var/run/bitlbee/ chown bitlbee:bitlbee /var/run/bitlbee/

helps to get rid of the:

Jan 15 15:40:39 icsserver.loc bitlbee[10722]: Warning: Unable to bind UNIX socket to /var/run/bitlbee/bitlbee.sock: No such file or directory Jan 15 15:40:39 icsserver.loc bitlbee[10722]: Unable to bind UNIX socket to /var/run/bitlbee/bitlbee.sock: No such file or directory Jan 15 15:40:39 icsserver.loc bitlbee[10722]: Warning: Warning: Couldn't write PID to `/var/run/bitlbee/bitlbee.pid' Jan 15 15:40:39 icsserver.loc bitlbee[10722]: Warning: Couldn't write PID to `/var/run/bitlbee/bitlbee.pid'

HOORAY FOR DX! Thank You for Your time for Your help and for Your patience for my stupidity...

Kind regards.

Andrzej

P.S. For Arch Linux users that want to use Bilbee with libpurple (patched so it retains gadu-gadu contacts) without X on their machine (minimal install of libpurple) here are my (well not mine as they were not written / created by myself but assembled from someone else work and only modified by me when needed) PKGBUILDS that will make the job of compiling / installing much easier...

​http://andrzejl.cyryl.net/AndrzejL/Files/PKGBUILD/libpurple-minimal.tar.gz ​http://andrzejl.cyryl.net/AndrzejL/Files/PKGBUILD/bitlbee-libpurple.tar.gz

Here is also Arch Wiki link that will help to setup Bitlbee correctly.

​https://wiki.archlinux.org/index.php/bitlbee

[dx]

The patch updated to 3.2.1

[robert@…]

Even the patch might work for Gadu-Gadu, it is unfortunately not generic enough. Just adding WhatsApp to that list in the patch makes BitlBee segfaulting after adding at least one WhatsApp contact, saving and reconnecting to WhatsApp. Given that WhatsApp (via libpurple) works fine in Pidgin and BitlBee without this patch, it brings the assumption that account_on() might be the wrong place (too early). On IRC dx raised the idea of using imcb_connected() instead while I wildly guessed about prplcb_conn_connected(). However my C is too bad to perform these changes even for testing purposes - however I am happy to test such changes if somebody provides me an experimental patch.

[dx]

Gadugadu patch applied in rev 1036. Whatsapp side of it, well, still pending. Sorry for that.

[dx]

here's the fix for the whatsapp side, will upstream it soon™ (and yes it turned out to be what i said about imcb_connected, yes i know it took too long, don't look at me that way)

[dx]

This stuff was applied months ago, forgot to close it.