This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: More bug fixing needed
- From: Carlos O'Donell <carlos at systemhalted dot org>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Thu, 30 Aug 2012 10:07:36 -0400
- Subject: Re: More bug fixing needed
- References: <Pine.LNX.4.64.1208301143200.9433@digraph.polyomino.org.uk>
On 8/30/2012 8:02 AM, Joseph S. Myers wrote:
> The number of open glibc bugs has been increasing over the past few
> months. There are various possible reasons for this:
>
> * People not closing bugs when they fix them. Patches have recently been
> checked in for bugs 14476 and 14516, for example, but those bugs are still
> open. If you've checked in patches for any bugs lately, and the patches
> completely fix the bugs (rather than only some cases), please make sure
> that you've marked those bugs as RESOLVED/FIXED, with reference to the
> relevant commit. The responsibilities of a patch committer include both
> adding fixed bug numbers to NEWS and closing the bugs themselves.
This would imply we are actually closing a lot of bugs, which we aren't IMO.
> * People being more willing now to file any bugs found, or cleanups where
> it's agreed that a change is desired and there is a clear well-defined
> completion criterion, in Bugzilla (rather than just in private todo list,
> or complaining about the bugs elsewhere without actually reporting them).
> This is good, but with more active bug reporters we need more active bug
> fixing.
This is part of the issue, and that's a good thing.
> * Many of the easier old bugs having been fixed in 2.16, so bugs that are
> left tend to be harder to fix.
This is true for quite a few of the bugs.
I find that glibc bugs are easy to find and hard to fix because of the
conservative nature of the project with regards to compatibility, and
that's just a burden we have to bear.
> * Development generally being less active at this time of year.
This is also true.
> However, ultimately we simply need more people fixing more bugs to get the
> number down. My guess is that probably no more than 100 of the 498 open
> bugs are actually hard in the sense of involving difficult design issues,
> difficult questions as to whether the requested change is actually
> desirable or more than a few hours' work to fix the bug. Bug fixing is a
> good way to learn more about and get involved in different areas of glibc.
> If everyone of the listed people with commit access were to fix 14 of the
> existing open bugs, they'd all be resolved (although quite a few of those
> people are inactive, I think some more people have had commit access added
> over the past few months without necessarily updating the MAINTAINERS wiki
> page, and you don't need commit access to fix bugs). If everyone with
> commit access were to fix just 3 of the existing open bugs, there would be
> fewer than 400 left, which would be clear progress.
I agree. Hopefully my review last night of Jeff Law's locale changes will
close out one more bug.
One thing we might do is have a bug triage day and time. Say Thursday afternoon
we get together on IRC at a particular time and do a group triage, talk about
the bugs and work through N bugs in 30 minutes.
I think this would help get us into the habit of doing triage.
Comments?
Cheers,
Carlos.