Sprint discussion and planing

[Mark Kretschmann]

On Mon, Jan 17, 2011 at 8:56 AM, Teo Mrnjavac <teo at kde.org> wrote:
> On Mon, Jan 17, 2011 at 07:10, Mark Kretschmann <kretschmann at kde.org> wrote:
>> On Thu, Jan 13, 2011 at 12:39 PM, Myriam Schweingruber <myriam at kde.org> wrote:
>> <snip>
>> * Rethinking our Git policies regarding rebasing and squashing
>
> If you think a discussion is needed then by all means let's discuss,
> but I don't see what's wrong with the current situation. I believe
> that the biggest disadvantage we have with the current policy (or
> absence of one) is that the occasional hothead drops by in #amarok to
> try to teach us the "right way".

There's nothing inherently wrong with our current policy, that is,
"there is no policy". I tend to think that everyone should work in a
way that matches their personal work flow best. However, there was one
recent incident that made me reconsider this a bit. It was this:

http://lists.kde.org/?l=amarok-devel&m=129440099506206&w=2

Rick has a work flow with Git which is basically making many small
commits. So far that is fine. He does not seem to squash a lot, but
rather pushes what he commits. The problem with that: Spotting one bad
commit in a stream of hundreds can be really hard. Now, Rick is a
great coder, he does not usually make mistakes. But still, it can
happen to all of us, and having a regression "sneak in" among many
commits that appear harmless, that's what I got afraid of.

So I would simply want to suggest that people with this style of
working should try to squash their commits, not radically, but grouped
into logical blocks. E.g. a patch that changes feature XY should go in
as one commit, but not 20. This makes it easier to keep an overview.

-- 
Mark Kretschmann
Amarok Developer, CEO of Kretschmann Software Consulting
Fellow of the Free Software Foundation Europe
http://amarok.kde.org - http://fsfe.org - http://kde.org