* remove EOL ruby versions for security and maintenance reasons.
* only expose ruby_MAJOR_MINOR to the top-level. we don't provide
guarantees for the TINY version.
* mark all related packages as broken
* switch the default ruby version from 2.3.x to 2.4.x
The patches do no longer apply on-top of the pjsip version we provice.
One of the maintainers (@Radvendii @olynch) should have a look what can
be done about this.
The build has been failing since 2017-07-31…
This version bump contains the following fixes:
v8:
- Avoid a segmentation fault when notmuch_message_get_header() returns NULL. (Patch by Víctor M. Jáquez <vjaquez@igalia.com>).
v9:
- Do not use the deprecated notmuch_query_count_messages_st function when using Notmuch 0.25. (Patch by Adam Ruzicka <a.ruzicka@outlook.com>.)
We already had this patch in our tree.
- Fix crash in when running queries and the program is build with Notmuch 0.25. (Patch by Joshua Krusell <joshua.krusell@v-dem.net>).
- Fix conditional compilation when using Notmuch 0.25. (Patch by David Bremner <bremner@debian.org>).
Adaptation of #23058 by @spinus to 1.0.0rc1. See #23058 and #28111 for discussion.
This patch keeps the warnings. But I think it is useful to have a sanely working
version in nixpkgs, even if it is still considered insecure, just to follow the
upstream development by actually running it instead of reading the log.
Upstream changes:
* Improve Zeroconf behavior
* Fix showing normal message event
* remove usage of OpenSSL.rand
* a few minor bugfixes
The really important part here is the third point about OpenSSL.rand,
because the rand attribute no longer exists in pyopenssl and thus Gajim
doesn't even start.
Also the fix-tests.patch has been fixed upstream as well, so we don't
need it anymore.
Another change in 0.16.9 that's not included in the changelog is that
there is a test_nogui target, which is also run by the CI upstream is
using, so let's use that and remove xvfb_run.
Signed-off-by: aszlig <aszlig@nix.build>
Cc: @7c6f434c, @Mic92
