Re: gothicx-guest removed
* Jon Dowland <lists@alcopop.org> [2007-10-29 12:30:33 CET]:
> I'm a little concerned that we're formalising stuff too
> much. How big a problem are bad commits to SVN? I've
> experienced one on prboom, and from reading the threads
> there has been at least one more to wesnoth, both by the
> same person.
And in wormux and in vodovod and in tennix and possibly in others,
too. Being able to say that they were by the same person is nice, but
technical it could had been more because that one person joined the same
way as most of the others.
> The commits in question have been caught and reverted.
Which was a lucky incident - I wouldn't have gone through looking for
other commits by him in other tree parts of the svn if it weren't for
his breakages and silly commits in the wesnoth part and his not really
satisfying responses to my questions on IRC.
> All the changes go to a public list which we are free to review.
... which doesn't seem to have been done, unfortunately. Thanks to
eddy's push^Winsight I now feel responsible to do so and subscribed the
list. *sigh* But that doesn't mean others can't/shouldn't, and even
though I strive for it I'm far from perfect myself so can easily
overlook parts, especially in parts that I'm not closely related to and
aren't obvious breakages like watchfile changes that simply can't match,
ever.
> Stuff in trunk does not automatically mean stuff in the
> archive. Presumably someone then checks out, adjusts
> changelog, performs some quality checks before getting their
> sponsor to upload, and the sponsor presumably either also
> performs some checks or has an established trust with the
> sponsee. (I don't have a regular sponsor so I'd ask
> games-team sponsors to weigh in here if this isn't accurate)
That's pretty much persuming and being sponsoring myself regularly
seeing sponsors as safeguard for silly commits is pretty ... not
fulfilling and not making you a trustworthy person in my opinion. I can
see why you say so, but it still shouldn't be the sponsor's job to
notice sillynesses, I highly doubt that any of them really would check
a watch file, for a start.
> Perhaps new people could be encouraged to work on branches
> rather than trunk?
Might also be a valid option, I don't know. Branches are much more
natural in distributed VCSes, but I am afraid they wouldn't work out for
the games team pool, it's too diverse, and splitting it up will
eliminate loads of the things we cherish.
So long,
Rhonda
Reply to: