This was submitted as a Self-Contained Change proposal, but it would seem to be more appropriately categorized as a System-Wide Change
Upgrade bash to 5.0 release. This release fixes several outstanding bugs in bash-4.4 and introduces several new features. The most significant bug fixes are an overhaul of how nameref variables resolve and a number of potential out-of-bounds memory errors discovered via fuzzing.
@frantisekz @adamwill What's the QA opinion here?
You can see some discussion of it in the logs from our most recent meeting, starting at 16:45.
tl;dr you are concerned but there is no conclusion whether it would be OK to get it in or not.
That didn't help much :D
well basically I'm pretty concerned, but everyone else seemed a bit less concerned, so there wasn't enough of a consensus for an official "QA Thinks This" pronouncement :D
Personally, I think it would be a lot saner to push this to F31 (and land it soon after F30 branching).
I see @adamwill's concerns, but I think they are overly pessimistic. The release notes imply that there are some changes, but those seem to be corner cases and shouldn't have a wide impact. If it turns out that this expectation was wrong, we can always revert in branched. So I think this should go in, and as usual in such cases, as quickly as possible.
+1
@zbyszek the problem is that just one "corner case" can have a very "wide impact" in a Linux distribution, if said corner cases happens to be in, oh, let's say...dracut. You know, that gigantic pile of shell scripts we have that produces our initramfs images...
Sure, but the same applies to many other packages, and we update those every release or every other release. glibc, gcc, the kernel, systemd, etc. Dracut is mostly written in fairly modern bash, so I don't expect it to break, but if it does, we have procedures in place to deal with that.
If it would make you feel any better, I'd be happy to generate some initramfses after the bash update and compare if they differ in any significant way. I need to to a diffoscope update anyway ;)
I guess I am a weak +1 on it. It does seem late and could be risky, but I think since it's out everyone will expect fedora 30 to be using it.
Metadata Update from @jforbes: - Issue tagged with: meeting
AGREED: F30 Change: Bash 5.0 is approved (+7,0,-0) (jforbes,15:42:53)
Metadata Update from @jforbes: - Issue close_status updated to: Accepted - Issue status updated to: Closed (was: Open)
Login to comment on this ticket.