hello,
On 05/07/13 19:55, David Kastrup wrote:
Mark Polesky <markpole...@yahoo.com> writes:
Should I start a new tracker issue for every little change I
want to make?
Yes. It's easy to do with git-cl.
That seems excessive.
It gives independent verification.
Should I just push to staging? That seems counter to the
spirit of the whole countdown thing.
The "countdown" thing and the "issue" thing is for the situation where
input is a possibility. Skipping either is a possibility, but if your
judgment about the non-necessity of feedback was faulty, expect raised
eyebrows. Pushing straightforward spelling fixes without markup
directly to staging is not much of an issue. For other things, our
mileage may vary.
Also note that a few of the devs simply don't have time to read through
all the emails or check a tracker on the off chance.
The email countdown thing we do, gives a clear and concise list of
patches that, if there are no undue comments, will be pushed. It also
means that some 'big' patches can be split up much more easily into
manageable chunks for easy review and comment.
Prior to this we had master breaking every week and more time was spent
finding the culprit, picking apart the problem, reverting and then
fixing it. It got quite dispiriting frankly.
The idea is that a dev makes a fix, uses git-cl and then pretty much has
nothing else to do, it gets fully tested (independently - usually
including a full make doc (when I test patches), which takes hours on
others' machines) and shepherded through.
Assuming that the patch is ok, we're talking a few days delay, max and
it gives ample opportunity for those that may (for example) only look at
the email lists a few days a week.
What's the rush?
James
_______________________________________________
lilypond-devel mailing list
lilypond-devel@gnu.org
https://lists.gnu.org/mailman/listinfo/lilypond-devel
