This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Policy for reverting someone else commit?
On Sun, May 20, 2018 at 07:27:06AM +0200, Richard Biener wrote:
> On May 20, 2018 7:20:25 AM GMT+02:00, Steve Kargl <sgk@troutmask.apl.washington.edu> wrote:
> >So, there is a P1 blocking bootstrap failure on trunk.
> >I've opened a PR and finally had time to locate the
> >offending commit.
> >
> >https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85843
> >
> >As I cannot bootstrap gcc, I cannot test a set of
> >patches for gfortran that I have in my tree nor
> >identify which recent commit introduced a regression
> >in the gfortran testsuite.
> >
> >I've scanned gcc.gnu.org and wiki, but have not
> >been able to find a stated policy of reverting a
> >patch committed by someone.
> >
> >The offending commit was done on a Friday. I
> >have no idea if the committer responsible for
> >the bootstrap failure works on the weekend.
> >
> >So, can I revert the commit (and don't in my
> >local repository)?
>
> IIRC there is a 24h rule that global maintainers can invoke.
> Not sure if that is formally documented somewhere.
>
> Usually it's much easier to revert this in your local repo
> for the time being.
>
Yes, I've reverted locally. I must be much more cautious
than others. I bootstrap gcc in a clean directory prior
to committing. Adding additional commits on top of a
known bad commit, which causes a bootstrap failure, seems
to be asking for trouble. I'll wait until tomorrow.
--
Steve
20170425 https://www.youtube.com/watch?v=VWUpyCsUKR4
20161221 https://www.youtube.com/watch?v=IbCHE-hONow