GitLab

There are small patches for libgpg-error and for libgcrypt

Thanks to Guo Yixuan <culu.gyx@gmail.com> for the report and patch.

Mostly in documentation, but also in the HELPURL in the code

Thanks to Daniel Kraft <d@domob.eu> for reporting the issue.

If the OTR plugin doesn't know about Groupwise's maximum message size,
then encrypted authentication messages will be undesirably truncated.

Libpurple declares Groupwise's maximum size as 2048.  That didn't work
for OTR, so we use 1792 - that one works.

https://bugzilla.novell.com/show_bug.cgi?id=809052

Thanks to Daniel Atallah <datallah@pidgin.im> for noticing that it
wasn't working before.

configure checks for the availability of i18n, but if it's not there,
we weren't doing anything sensible.  Now, we just use the English
constant text strings in the code by defining _(x) and N_(x) to be just
(x).

When switching to a conversation tab for which the user has explicitly
disabled OTR, remove the OTR menus from the top menu bar.  Thanks to
Greg Troxel <gdt@ir.bbn.com> for reporting the issue.

It used to say "4.0.0 or newer"; now it says "4.x >= 4.0.0", since 5.0
is *not* acceptable.  Thanks to Jacob Appelbaum <jacob@appelbaum.net>
for pointing it out.

If one of our buddies is/her side of the OTR session, don't
automatically change bel on the OTR button on our side to "Finished",
since may be other sessions for the same buddy still in ENCRYPTED.
This was the last call to dialog_update_label_conv that still had a
hardcoded trust label set, rather than using
otrg_plugin_context_to_trust to get the trust label corresponding to the
currently selected instance.

When we're looking up information from
purple_conversation_get_data(conv, ...), it may happen that the result
is NULL because pidgin has emitted conversation-switched before emitting
conversation-created, so we try to switch to a conversation that the
conversation-created callback hasn't yet filled in the data for.  This
causes a crash.  Be more defensive about this.

We should only free the userstate after all the pidgin stuff is cleaned
up, as some of it may use the context.  Indeed, this fixes a crash bug
when you try to unload the OTR module while a SMP session is in
progress.

The right thing to do is to deallocate things in the reverse order that
you allocated them, which is what we do now.

We had some calls to gtk_box_pack_start() that were trying to pack
already-packed widgets, yielding:

Gtk: gtk_box_pack: assertion `child->parent == NULL' failed

clist_all_unselected() was marking the four buttons in the OTR
"Known fingerprints" configuration as unselectable, but sometimes before
they existed.  Check for existence first.

Ensure "-g -O2" is also in the CFLAGS (in addition to the hardening
options) when building libgpg-error and libgcrypt on win32.

Change all version numbers to remove "-beta2" and turn off the beta
warning dialog.

Add -fno-strict-overflow -Wall -Wextra -Wno-unused-parameter
    -Wno-missing-field-initializers -Wformat-security