This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Fix bootstrap (PR rtl-opt/29879)


On Mon, 2006-11-20 at 16:44 -0500, Richard Kenner wrote:
> > Agreed.  We don't need special rules here.  This is just common sense. 
> > Checking in a major patch that is more than a few hours old without 
> > doing another full bootstrap/test cycle is simply not acceptable.
> 
> We have to be careful here.  Not all GCC developers have access to fast
> enough machines that a full bootstrap/test cycle can be done within "a few
> hours".  For some people, it's still a "start the run, go to sleep, and check
> the results in the morning" situation.  I don't think it benefits the GCC
> project to exclude those people from contributing.
> 
> Obviously, we want the "update tree, test, and commit" process to be as
> atomic as possible, but I think "a few hours" is unreasonably optimistic
> for many people.
> 

Or whatever it takes. If it takes 10 hours, fine. I'm not trying to
exclude anyone, but 2 weeks is excessive by any stretch of the
imagination.

Even a source tree update/incremental bootstrap to ensure bootstrap
still works on a slow machine should be minimally possible.  And even
then, as long as the testsuite run continues after the checkin to ensure
no problems were encountered, that works too. 

Im just trying to say the person doing the checking-in needs to be
responsible to make "best effort" subject to the limitations of their
equipment to ensure the patch works with the branch at checkin time. 

Andrew


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]