[Python-Dev] Checked in...
Guido van Rossum
guido@beopen.com
Sun, 16 Jul 2000 18:53:41 -0500
[Greg]
> To defend Paul here a bit, we did start some discussion here, and at least
> some of us said "the concept is fine, so check it in. we can totally revamp
> as necessary."
I can't even go away for a weekend and they start checking stuff in
behind my back. ;-)
> CVS is a much better communication/testing vehicle than the damn Patch
> Manager or posting modules at some private web site.
>
> And we can always "cvs rm" a module.
Actually, this leaves a turd in the Attic directory.
Adding something to the CVS repository gives it a false legitimacy --
not everybody reads the "experimental feature" comment.
> Another alternative would be to create /nondist/sandbox/ and have people
> toss modules-for-discussion or modules-in-progress into that area. We can
> dork around as much as needed. When the module is "ready", then we add it
> into the /dist/ branch.
I like this much better!
> CVS is a great tool for this. Let's not try to create roadblocks that
> prevent effective work.
Sure. But let's be careful with what we check in, especially into the
"dist" source tree. This is not a place for random experiments.
I've seen a few examples recently where it seemed some kind of
discussion was being carried out by checkins. Regardless of who's
right or wrong, that's the wrong medium for such a process.
[Tim]
> There was a discussion, overwhelmingly positive, and then to make further
> progress Paul asked here whether it wouldn't be easier to just check
> something in so we could all kick the tires. At least three of us (four if
> you count me twice, as I tend to do <wink>) said "sure!".
>
> > ...
> > Annoyed,
>
> I can understand that too <wink>. But as Paul said, it's a self-contained
> module that "can't" break anything else, and he did ask first, so maybe you
> could look on it a bit more kindly this once.
Sure. I recommend the use of the nondist CVS tree for experiments.
But to me the dist subtree is sacred and should only contain code we
believe we agree on.
--Guido van Rossum (home page: http://dinsdale.python.org/~guido/)