Home
release-team@conference.openafs.org
Friday, February 7, 2020< ^ >
meffie has set the subject to: openafs release team
Room Configuration
Room Occupants

GMT+0
[12:05:21] mbarbosa joins the room
[16:53:48] mbarbosa leaves the room
[17:00:27] wiesand joins the room
[17:00:50] <wiesand> good evening
[17:02:09] <wiesand> hmm, "meffie has set the subject…" but I don't see "meffie" in the list of participants
[17:03:40] meffie joins the room
[17:03:52] <wiesand> ah, hello Mike
[17:05:07] <wiesand> I hope Ben is asleep and recovering.
[17:05:22] <wiesand> I don't have much on the agenda for today anyway.
[17:05:33] <meffie> hello, sorry, disconnected. i'm back.
[17:06:16] <meffie> the topic scroll back is just from when I set it some weeks ago.
[17:06:47] <wiesand> Thanks a lot for this week's review activity. It should enable me to make progress on 1.8.6pre1 over the next days.
[17:06:54] meffie has set the subject to: openafs 1.8.6pre1 preparation
[17:07:06] <meffie> (just do /topic to change the topic)
[17:07:33] <wiesand> (took a few days off early next week, that should help getting done something)
[17:08:11] wiesand has set the subject to: weekly openafs release team meeting
[17:08:34] <meffie> cheyenne pointed out 3 commits for 1.8.x to fix gcc warnings which were not on your list last week.
[17:08:52] <meffie> i can ask for the numbers if you like.
[17:09:07] <wiesand> that wouldn't harm, but
[17:09:53] <wiesand> chances are they have sufficient review IMO but are part of stacks with other changes which haven't
[17:12:42] <meffie> ok, i'll ask him again for the numbers.
[17:13:38] <meffie> 13740, 13732, 13731.  Original master commits were authored by Ben
[17:13:44] <wiesand> also tell him he's absolutely welcome to contact me directly when he has concerns I might be missing something
[17:14:21] <meffie> thank you wiesand
[17:14:54] <wiesand> 13740: seems fine, but path conflicts with 13895 which did need review last week
[17:15:23] <wiesand> 13732: the same, with 13897
[17:15:49] <wiesand> 13731: the same, with 13896
[17:17:05] <wiesand> so they're on my watch list - I just wanted to keep the list of changes I asked to be reviewed to those who actually required it
[17:17:22] <wiesand> (all IMHO of course)
[17:18:16] <meffie> how do you see which ones it "path conflicts with", for example 13740?
[17:18:44] <wiesand> you don't want to know (though you've seen it before)
[17:19:50] <meffie> those commits dont conflict tho?
[17:20:05] <meffie> the change different parts of the same file.
[17:20:15] <meffie> s/the/they/
[17:20:25] <wiesand> I keep a list of which change touches which files - if two changes touch the same file, they path conflict unless one is based upon the other.
[17:21:00] <wiesand> Yes, not a conflict. Still a "path conflict" for gerrit.
[17:22:00] <meffie> well, i'm not sure if that is useful metric. you could have changes it two different files that causes a bug. and that's not a so called "path conflict"
[17:22:47] <wiesand> I'll rebase pretty much everything anyway, and I don't expect problems. Yet, if you miss those path conflicts when rebasing, you'll get to do it again, at least partly.
[17:23:39] <meffie> ok, i guess this helps keep a clean history. thank you.
[17:24:37] <wiesand> It's not perfect. But it's what you can expect gerrit to do. It's a conservative setting chosen when git/gerrit for openafs were set up long ago, and I fully agree with it.
[17:24:55] <wiesand> No obscure merges.
[17:25:34] <wiesand> The case you mentioned, well that's our job to prevent.
[17:26:32] <meffie> i see, the "path conflicts" now, on the top right, next to related changes. i'll look out for that from now on.
[17:27:03] <meffie> and let the other know too.
[17:27:44] <meffie> i will let them know so they can behave with the commits.
[17:28:28] <wiesand> it's "very easy": whenever you pull up a change, simply check all others already pulled up whether they touch any common file, and base your pullup on that one
[17:29:09] e632fa5c joins the room
[17:29:13] <meffie> yes, only if someone knows to do that. i did not realize such.
[17:29:46] e632fa5c leaves the room
[17:29:53] <meffie> i normally just cherry-pick on top of origin/openafs-stable-1_8_x
[17:30:13] e632fa5c joins the room
[17:30:27] <wiesand> alternatively, hold your horses - bring up the changes that should be pulled up and let me handle that, or wait until the backlog is shrunk by merging stuff away
[17:30:45] <wiesand> yes, everybody does, and that's the problem ;-)
[17:30:57] <meffie> we did not know.
[17:32:06] <meffie> should we have a different "staging" branch in gerrit, then i can just rebase those all before pushing to your openafs-stable-1_8_x branch?
[17:32:17] <wiesand> I tried to point it out before - obviously not too successfully ;-)
[17:32:46] <meffie> sorry i did not understand.
[17:33:19] <wiesand> The best way to tackle this is to not allow the backlog to grow that much.
[17:33:50] <meffie> yeah, but we want to have a place to put candidates for stable.
[17:34:10] <meffie> maybe we do not?
[17:35:37] <meffie> i guess it would still make a backlog. but maybe we can sort it out before commits are pushed to gerrit.
[17:36:49] <meffie> maybe there is way to have gerrit reject commits that have a "path conflict"? i need to think about this.
[17:39:34] <wiesand> then your super-important pullup of the day will be rejected because someone else pushed a change fixing a typo in a completely unrelated part of the file
[17:40:00] <meffie> oh. hmm.
[17:40:03] mbarbosa joins the room
[17:40:18] e632fa5c leaves the room
[17:40:20] cwills joins the room
[17:41:43] <meffie> ok, i guess the point is to avoid path conflicts when pushing to the "openafs-stabe-1_8_x" branch. that should help. i'll update the wiki to make that more clean, and give some examples. i'll need to update git-gerrit cherry-pick too.
[17:42:29] <wiesand> the work required to get rid of the backlog grows quadratically with its size
[17:42:34] <meffie> sorry for being slow to learn!
[17:42:52] <meffie> yes.
[17:43:28] <wiesand> and at 1h buildbot turnaround time per rebased change, that's true for the wall clock time too
[17:43:40] <meffie> ok
[17:44:20] <meffie> sorry for getting off topic.
[17:44:33] <wiesand> really, if there are 70 changes pulled up already, no point in pulling up more - keep a list of your pets and invest your time in getting the open ones sorted out (review review review)
[17:45:37] <meffie> is there a way you can turn on or off the gerrit branch?
[17:46:11] <meffie> eg. "the inn is full", you will need to wait for the next regularly scheduled bus.
[17:46:38] <wiesand> no, I can -2 everything new coming in, unless crtitical, though - and I've done it in the past
[17:47:19] <wiesand> (just like Ben before cutting 1.8.x)
[17:47:42] <meffie> does it make sense to announce on the mail list when you would like to accept gerrits?
[17:48:29] <meffie> so the backlog would to grow and you would not need to mark -2?
[17:49:03] <wiesand> as you noted, that will just move the backlog somewhere else ;-)
[17:49:15] <meffie> lol, yes.
[17:49:36] <wiesand> the only solution, really, is to keep the backlog short
[17:50:29] <meffie> i see. at least we can try to avoid path conflicts when possible.
[17:50:49] <wiesand> this week, for example, reviewing open changes was much more likely to accelerate getting your favourite change of the week in than pulling it up
[17:51:29] <meffie> we used the list you posted last week to focus on the ones you requested.
[17:52:09] <meffie> i hope that was helpful.
[17:52:52] <meffie> again, sorry for straying from the topic.
[17:53:29] <meffie> i am just cranky and hungry :) this meeting now replaces lunch.
[17:53:53] <wiesand> I saw that, and appreciate it very much! It was very helpful. Especially if now I manage to get at least many of those merged during the next days.
[17:54:20] <wiesand> No, I'm glad to discuss these issues.
[17:54:27] <meffie> ok thank you.
[17:56:33] <meffie> are there more gerrit numbers we should review for openafs-stable-1_8_x?
[17:57:08] <meffie> otherwise, i will just work on the master branch ones for the next few days.
[17:58:09] <wiesand> I have to have a closer look, but chances are I'll be able to make significant progress thanks to the reviews done this week.
[17:59:03] <meffie> very good. may i motion to adjourn?
[17:59:13] <wiesand> Go get something to eat ;-)
[17:59:26] <wiesand> Thanks a lot Mike!
[17:59:50] <meffie> thanks! oh, i forgot to mention, the windows builders should be working faster now. let me know if you see a problem.
[18:00:01] <meffie> have a good weekend!
[18:00:14] <wiesand> great news! I'll need them a lot soon.
[18:00:20] meffie leaves the room
[18:00:25] wiesand leaves the room
[18:01:54] cwills leaves the room
[19:10:36] mbarbosa leaves the room
[19:41:58] mbarbosa joins the room
[21:46:47] mbarbosa leaves the room