[Kde-scm-interest] the permissions confusion

[Oswald Buddenhagen]

On Sat, Dec 12, 2009 at 04:58:56PM +0100, Cornelius Schumacher wrote:
> On Saturday 12 December 2009 Oswald Buddenhagen wrote:
> > yes, it's great to live in a filtered reality. i suggest you browse
> > the kde-commits archives and check *how* often somebody complains
> > about violations of project policy.
> 
> That's actually an interesting question. It's a bit hard to do that,
> though, because there are no obvious search terms.
> 
yes, i've been wondering how to do that, too. :)
first, one has to determine the "usual operation areas" for all
committers, and filter out all threads which relate to these. this
leaves the threads where developers "stray out", which is really what we
are interested in currently. i expect this to be well under 1% of all
commits in the first place, which still means a few a day. making a
survey on that data should be possible (and probably necessary) without
techical means.

of course, that's quite an overkill *), so i'll just rely on my gut
feeling which tells me that it happens too often.

*) actually, finding these "operation areas" plays into the (totally
off-topic ;) "maintenance coverage" idea raised elsewhere, so it
wouldn't be entirely wasted effort. :)

> Call me non-paranoid, but I wouldn't bother with stuff which happens
> this rarely.
> 
just because you aren't paranoid doesn't mean that they are not after
you ... :D

anyway, besides preventing the extreme case where somebody abuses his
commit rights (be it intentionally or not) and somebody gets upset about
it, such a system (i mean the magic push) would foster a culture of
review, as it would be less of a burden to do it.

installing such a system also doesn't preclude allowing overrides for
urgent cases (say, obvious build breakage and no maintainer is to be
found on IRC). at qtdf, we have GIT_FORCE=yes-please to enable forced
pushes in justified cases. i'd generalize that to
GIT_OVERRIDE={acl,force,...}.

all that being said, this isn't something which needs to be done right
now. i'm just elaborating on the "mostly". ;)