Hello Diedrrik.
Thank you for your hard work here, I appreciate it. I'm quite
overwhelmed with other immediate responsibilities recently. I'm bringing
my laptop on a trip I'm about to take and I might have time to do some
work on the package, but I don't think I'll be able to do much with this
bug.
The good news is that the Mumble package can be configured not to use or
compile with Ice and I've done that, so at least the package can remain
in Debian while upstream can figure out how to deal with it.
In the past Mumble upstream has had me release a Mumble version from
Git, which is at least possible, if it comes down to that when the
Debian release approaches. Last I did this years ago it required making
a script to put together several sources and remove unreleasable files
to be able to build a releasable tarball, which then had to be released
without any cryptographic signature. Hopefully we won't have to resort
to that.
I'm also not really familiar with Zeroc-Ice and I don't use the features
it provides the Mumble package. It's used to control and automate
mumble-server operations, so it's "advanced features" mainly for sites
that operate large mumble-server installations that have a lot of users
and moderators. For those users hopefully they can hold back their
Mumble-server version to continue to use Ice until this can be fixed.
For less advanced users that only need a simple mumble-server instance
the newer versions without Ice support will still function fine.
The only other thing I can mention here is that the recent Debian GR
concerning LLM support came down to a decision that AI use is acceptable
in Debian via "responsible use", meaning any uploads done are still the
responsibility of the uploader, and AI use does not need to be
disclosed. So if you want to make progress and get to a point where all
seems lost or get stuck, it is acceptable to ask an AI (especially
locally via LM-Studio GUI or other "AI collar" to keep the AI within its
desired context) to see if a hint or a solution can be found to make
progress. It's a tool and useful for some things, but it can also
hallucinate and "pull a Star Trek and make up technobabble", so the
results need checking if it seems to work. But "when all else fails"
it's worth a shot.
Thanks again and I hope things are well with you.
-- Chris
--
Chris Knadle
[email protected]
On 10/7/26 2:16 PM, Diederik de Haas wrote:
On Sun Sep 27, 2026 at 9:21 PM CEST, Diederik de Haas wrote:
On 9/12/26 10:35 AM, Diederik de Haas wrote:
In upstream issue #7067 one of the mumble maintainers asked for testing of
https://github.com/mumble-voip/mumble/pull/7290
https://github.com/mumble-voip/mumble/issues/7067#issuecomment-5297216501 :
"Could someone please test the changes in #7290 against a version of Ice >=
3.8? I currently don't have such a version at hand for testing so I can't do it
on my own (easily)."
I've made a serious attempt and I've gotten further, but the build still fails
and fixing that is beyond my capabilities.
I've been trying to make it compile against 1.5 at all, but there's a high
probability I'm doing it wrong and/or have 'unforseen consequences'.
And the changes/backports I've made thus far would be unsuitable for a Stable
release because they're way too invasive.
I (also) know nothing about zeroc-ice, so I have no idea how to test it to
see if it actually works properly ... if I could get it to compile.
https://github.com/mumble-voip/mumble/issues/7067#issuecomment-5875675288
"Yeah, 1.5 is still stable but the fix is targeted at master and
won't be backported to 1.5 (officially)"
As upstream has no intention to make a backport to 1.5 (which I can
understand), I'd say mumble 1.5 cannot have Ice integration.
Cheers,
Diederik
