This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: No bug or public post for commit bc680b336971305cb39896b30d72dc7101b62242
On 08/27/2018 02:46 PM, Paul Eggert wrote:
> Carlos O'Donell wrote:
>> I don't see a public bug or post for commit
>> bc680b336971305cb39896b30d72dc7101b62242, did they get stuck or
>> miss getting sent?
>
> All my recent regex changes have been for Gnulib only and have no
> effect on glibc as far as I know. Although the commit you mention
> affects code outside the #ifndef _LIBC area, Assaf's email
> <https://lists.gnu.org/r/bug-gnulib/2018-08/msg00071.html> indicated
> that the bug occurs only when _REGEX_LARGE_OFFSETS is 1, which is
> true only in Gnulib.
An email with [committed] with exactly this message would have been
great, and would have avoided any confusion over what you were
working on, or if you had accidentally committed something without
meaning to do so.
> If after considering this you still think it's necessary to go
> through the glibc review process for patches that affect only Gnulib,
> I can do that. However, my own feeling is that the overall cost of
> this bureaucracy would exceed its benefit for Gnulib-only changes.
Yes, you still have to go through some of the bureaucracy. At the very
least a post to libc-alpha with [committed] is required. Joseph Myers
sets the best example among all of us.
Regarding bugs, they identify defects and provide downstream with an
audit trail, but if you didn't fix a glibc bug then you don't need
one.
The bureaucracy is an important part of the expectations we have set
with downstream. At least in Fedora and RHEL we use the bugs to actively
track which things need fixing, and it has been very useful that the
community adopted these particular strategies.
Consensus so far is:
* Post your patch on libc-alpha.
* If you fix a publicly visible bug, create a bug to track it.
> For what it's worth, as far as I can tell no current glibc
> contributor understands the regex code; its original author has not
> replied to my emails. That being said, I probably understand the it
> better than any other current contributor.
Raise your hand and become a subsystem maintainer! :-)
--
Cheers,
Carlos.