Am 21.12.22 um 18:21 schrieb Manfred Hollstein:
On Wed, 21 Dec 2022, 17:43:18 +0100, John Paul Adrian Glaubitz wrote:
On 12/21/22 17:37, Manfred Hollstein wrote:
So I can import that patch as well meanwhile. But it's always a timing issue. I want to submit FF on release day and very often it's not possible to get a fix in time.
It's always a good idea to look at Debian unstable when it comes to Firefox. The maintainer there, Mike Homney, is also a Firefox developer and Mozilla employee. So, he usually has an idea how to fix such issues.
to be honest, that's not a fix! As Wolfgang pointed out, such failures should be identified and fixed during the release phase. Adding this as a patch after release doesn't smell too good to me...
I haven't looked at Mike's particular fix for the i386 issue.
What's the problem?
as I wrote in my second e-mail, patches like this would be a perfect candidate to be included in another minor release such as 108.1 - which has been released in the meantime. If he is a Mozilla employee and a Firefox developer, he should have pushed such a patch long ago *before* the release - but that's just my opinion...
Just to address a few points from the conversation:
I check Fedora and Debian repos once in a while. I also sometimes chat with Mike and Martin as I know both from personal meetings from the past. The main point is as I pointed out the timing and apparently mozilla's way to handle such build issues. I cannot tell why those are only found after a release or if they are found earlier, why they are not addressed directly upstream before a release. I can only assume that upstream does not care much about some architectures.
And yes, part of the problem is also that I'm constrained in time. Firefox releases are happening every 4-5 weeks and 90% are failing to build in less common architectures. If my daily work would be to keep Firefox building along it certainly would be easier since I would be able to identify those issues sooner by building the beta tree every day during those 4 weeks.
In any case: - Dimstar's post here was a reaction on me excluding i386 from the build - from my side that was primarily because I don't see why a massive amount of users should be blocked from getting (security) updates over an undefined period of time until i386 gets eventually fixed - while I see good intentions in the reply, reacting with "do this and that" basically trying to teach me how to do my work and telling me to spend even more time feels a bit like not hitting the point - the i386 issue is fixed, the ppc64le one is not (yet)