On Mon, Feb 27, 2012 at 3:36 PM, Roland McGrath <roland@...> wrote:
>> There's also the related question when committing someone else's patch of
>> whether the submitter should be presumed to have tested the patch or
>> whether the committer needs to retest it.
> Committing for someone else is a courtesy toward those who can't commit
> themselves. The responsibility for the change rests with the submitter.
> The committer takes on a little responsibility for being fairly sure that
> the submitter did everything they should have, since they can't fix it up
> or revert it themselves either. Hence it is appropriate for the committer
> to make doubly sure by asking for explicit claims about testing and such,
> rather than presuming anything. At the end of the day, only the people
> with commit rights are really expected to be completely cognizant of and
> competent in all the procedures and conventions. But I think the main
> responsibility for actually doing the testing still lies with the submitter.
As a committer you don't need to retest anything, but it is absolutely
your right and responsibility to ask the tester if they did test and
under what conditions e.g. x86 and no regressions.