Hello,
On 22/01/2022 22:12, Carl Sorensen wrote:
James,
How much time do you spend per week as Patch Meister?
I spend about 15-20 mins every 2 days. I don't review code, I just check
reg tests (pasting those that are new or changed significantly) I also
make sure all threads that are still unresolved are either really
unresolved or just not closed (in the latter case I don't close threads
myself but will comment to make sure that the relevant parties know
there is a thread that looks like it is still TBD). I don't send
personal emails but do everything via GitLab so that all can see responses.
The original cadence was 4 days but that was when we had far fewer
active developers. It has only moved to 2 days because of the regularity
of MRs and patches.
I sometimes deliberately make the cadence longer if there are many
patches in the three queues so that everyone gets a chance.
So Log in, look at the MRs in countdown (and change them to push / leave
them or move them down, i.e. to review) then I look at MRs in review and
move them to countdown (or if there is still ongoig discussion I will
leave it there) and finally I look at MRs in 'new' and check reg tests
and set them to review.
So I start at the top so to speak and work down. (Coutdown/review/new).
Pushing everything up one step and out to patch Push.
I then use the 'countdown.py' script in the infrastructure repo that
generates the list you see in the emails. I add the date of the
countdown in the subject of the email and change the 'next countdown
date' in the body and then just send the email to dev.
My general rule is any patch in review that is less than 24 hours old
stays there (i.e. doesn't get moved on). I rely on developers that
resubmit patches, depending on where it was in the queue to either set
it back to the new/review queue or state in a note why they don;t think
it needs to start back at review again.
I try to avoid the 'this is trivial we should push this now' comment
because after all unless it is something that is affecting release, it's
only 48 hours before the next countdown. Some developers tend to be
impatient and I just ilke to make sure everyone gets to comment.
No comment in a thread is the same as a LGTM. This saves me having to
wait for people to approve explicitly every MR.
Sometimes I see a thread in a MR that is not going anywhere and no one
is replying or making a decision, so I will comment in the thread that
something must be done and if after the next countdown nothing has
changed I will 'do something' (be it set it to push or set it to needs
work) as that usually generates some attention. But some MRs have been
known to sit in 'review' for multiple countdowns and even at 'countdown'
until it is obvious that something needs to change. I just state why it
has not been moved (reference a comment for example) and then teh
developer can tell me why I am wrong or why X needs to wait or why Y can
be ignored. I always defer to the developers.
It's more or less just shuffling papers and checking reg tests with a
bit of comment and running the countdown script that I paste into the email.
Hope that helps.
James
If the time requirement is manageable, I could step in. I don't have lots of
time to spend on Lilypond (hence my lack of current development work), but I
could plan to spend some.
Please help me understand the time you spend.
Carl
On 1/22/22, 11:09 AM, "lilypond-devel on behalf of Jonas Hahnfeld via Discussions on
LilyPond development" <lilypond-devel-bounces+carl.d.sorensen+digest=gmail....@gnu.org
on behalf of lilypond-devel@gnu.org> wrote:
Am Donnerstag, dem 13.01.2022 um 22:42 +0100 schrieb David Kastrup:
> Kevin Barry <barr...@gmail.com> writes:
>
> > Hi James,
> >
> > This is a big loss; you will be missed. Thank you for all of your hard
work.
> >
> > > I want to stay a
> > > developer and that conflicts both time-wise and with respect to
> > > impartiality.
> >
> > Finding a non-developer to fill James's shoes sounds difficult. I
> > would volunteer, but I still plan to contribute patches.
>
> git shortlog --author Lowe
>
> should be some indication that this is not a conflict of interest as
> such. It may end up a conflict of resources.
Yes, my initial message wasn't entirely clear, it's of course possible
to contribute patches while being "Patch Meister". The "impartiality"
aspect is that you should try to apply the same standards to your own
submissions as to all other patches.
So, there's around one week left until the end of January and this
thread has more or less died down, thus last call for volunteers.
Jonas