[darcs-users] managing change

Gwern Branwen gwern0 at gmail.com
Mon Feb 9 02:10:00 UTC 2009


On Sun, Feb 8, 2009 at 8:48 PM, zooko <zooko at zooko.com> wrote:
>> Let's not forget the context of this. When it started this policy, twisted
>> was already a very successful and widely used framework. In other words it
>> had a big momentum.
>
> True.
>
>> Now this policy would be a completely different story if applied to a new
>> project you just started. I can guarantee that you will be the only
>> contributor to your project for a very, very long time. Only when it will
>> gain enough momentum, will people start to overcome the restriction and
>> contribute.
>
> Well, on the tahoe project (which is much smaller and newer), we've had
> about a half-dozen people offer patches.  Most of them were tiny bugfixes or
> tiny added options, but a couple were a bit larger.  In each case we said
> "Looks good, please provide thorough tests", and in every case the
> contributor got back to us (I think in all but on case in less than 24
> hours) with tests.
>
> So your hyperbole there about being the only contributor for a very, very
> long time is definitely not true in general.  :-)
>
> Really, are we sure that there are *any* potential contributors who are
> turned off by testing requirements?  I am well aware of some Twisted
> contributors who were turned off, but that wasn't because of the testing
> requirement, that was because of the *review* requirement, because nobody
> would get around to reviewing your patch and it wouldn't go anywhere.  That
> sort of open-ended delay is certainly a major turn-off.  I'm not sure that
> I'm really aware of any contributor who decided not to contribute to
> something because of testing standards.  It could well be seen as
> encouraging -- at least you know if the project has standards then your
> patch won't be broken by the next patch to land in trunk after it.  ;-)
>
> Regards,
>
> Zooko

I think what would be nice is if the Darcs cabal got together and  decided to:
1) Make official policy 'patches must have tests'
2) Secretly agree amongst themselves that patches rejected for lacking
tests would languish a week at most, and if the submitter doesn't come
up with tests, one of the cabalists would write tests for it.

Of course, this will never happen. TINC, y'know.

-- 
gwern


More information about the darcs-users mailing list