[phobos] [D-Programming-Language/phobos] a15e68: Accidentally committed with a bunch of debugging c...

[Lars Tandle Kyllingstad]

On Mon, 2011-05-02 at 17:05 -0400, David Simcha wrote:
> 
> 
> On Mon, May 2, 2011 at 4:45 PM, Jonathan M Davis <jmdavisProg at gmx.com>
> wrote:
>         
>         
>         I'm a bit divided on it. Andrei and Lars were pushing for
>         _all_ check-ins to
>         be reviewed. Part of me thinks that that's overkill and yet
>         there have been
>         times that other developers have caught stuff that I likely
>         wouldn't have even
>         with smaller changes.
>         
>         It does seem like overkill to require a pull request for
>         smaller changes,
>         particularly if they really don't look like they'd cause a
>         problem, but at the
>         same time, the extra eyes can be really valuable, even when
>         you don't expect
>         it.
>         
>         The truly minor, non-code stuff - such as updating the
>         changelog - shouldn't
>         need a pull request, but at this point, I'm inclined to agree
>         with Andrei and
>         Lars that all (or at least very nearly all) code changes
>         should go through
>         pull requests.
> 
> Actually, one of my big concerns with using pull requests is that I
> want near-instant feedback from the auto tester to make sure my stuff
> works on all platforms.  It's frustrating to have to wait an
> indeterminate amount of time for such feedback.

One option would be to commit all changes to a "testing" branch and have
the auto tester run off that.  Then, every once in a while, Andrei (or
someone else) can merge the testing branch with master, provided the
auto tester has succeeded for all platforms.

-Lars