[Pkg-silc-devel] Current state of SILC in Debian

[Micah Anderson]

Hi all,

After having yet another segfault in my irssi due to the silc plugin, I
decided that it might be a good idea to look at getting the debugging
symbols setup so that I can report proper backtraces. In doing so, I
realized that the versions that we have in Debian are outdated, and that
there was various things that needed to be done in the Debian SILC
packaging team. Here is the current state of things, as far as my most
recent efforts:

First, I reported #501466 against irssi to request that the patch to fix
the null dereference in skip_target() causing the silc plugin to
segfault. There is also a segfault issue in irssi not properly handling
overlapping keyboard_entry_redirect calls correctly[0] which has to be
handled in the irssi core. Sadly, the debugging work around this has not
received a response from the irssi developers (the original message was
posted in July!), I've posted a ping to the list to see if it was lost.

I updated our silc-client sources to 1.1.5 and updated the silc-server
sources to 1.1.3. In doing so I found that the silc-client sources had
mismatched sha256sums reported on their website. Turns out that they are
advertising sha256, but its actually sha512, I've sent the silc-devs a
message about correcting this.

Resolving the embedded source copies[1] issue needs to be taken care of,
but I dont think we should let that block us from uploading these newer
versions of the source as they resolve a number of crashing problems (as
well as some Lintian problems that dkg fixed).

However, as Jérémy noted[2], silc-client does use an embedded
copy of the silc library to build irssi-plugin-silc, and if we do
decide to build and upload a new package with these newer
sources, we should pull in those silc-toolkit fixes. I'm not sure
I understand what is involved here, and as a result not
comfortable doing this. 

Jérémy, perhaps you could describe what is involved here?

There has been a suggested fix in the BTS for the duplicate emote
messages (#476177) which would be great to attempt to bring in, as this
is pretty irritating. There are also a number of bugs that have been
reported that would be good to summarize and take to the silc-dev
mailing list.

On a more logisitical note, it was proposed, back in February[3], to
move our repository from subversion to git. I think that this would be a
good idea, especially now that upstream has also switched to git, making
collaboration much easier.

Finally, I requested GMANE to gate our pkg-silc-devel and
pkg-silc-commits lists as I find this service very useful and would
prefer to read things through there.

Micah

0. http://permalink.gmane.org/gmane.network.silc.devel/923
1. http://lists.alioth.debian.org/pipermail/pkg-silc-devel/2008-September/000349.html
2. http://lists.alioth.debian.org/pipermail/pkg-silc-devel/2008-July/000322.html
3. http://lists.alioth.debian.org/pipermail/pkg-silc-devel/2008-February/000239.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/pkg-silc-devel/attachments/20081007/4c56bce4/attachment.pgp