- Friday, January 4, 2019

release-team@conference.openafs.org

Friday, January 4, 2019< ^ >

Room Configuration

Room Occupants

GMT+0
[00:20:03] mvita leaves the room
[08:15:43] mvita joins the room
[09:20:14] mvita leaves the room
[13:55:45] meffie joins the room
[14:00:35] <meffie> happy new year.
[14:01:36] wiesand joins the room
[14:01:53] <wiesand> happy new year
[14:02:14] <wiesand> sorry for not showing up last week
[14:03:46] <meffie> no problem, it was a holiday week.
[14:04:37] mvita joins the room
[14:04:48] <mvita> hi
[14:05:12] <wiesand> Hi Mark
[14:05:45] <mvita> Mike said you missed me <3
[14:05:48] <wiesand> I just pulled up the "vos status" change - 13422
[14:06:03] <wiesand> Yes I did :)
[14:06:04] <meffie> thanks
[14:07:06] <wiesand> that one isn't for 1.6, right?
[14:08:14] <meffie> right
[14:08:25] <meffie> it's a regression in 1.8.x
[14:09:11] <mvita> oh, stephane, I had already pulled it up for you yesterday, 13421
[14:10:54] <wiesand> stupid me - abandoned
[14:11:51] <wiesand> thanks a lot for the 1.6 linux 4.20 changes
[14:12:01] <mvita> np
[14:12:06] <wiesand> I haven't looked at those yet
[14:12:25] <wiesand> I'm even unsure which ones we'll actually want
[14:12:38] <mvita> they were a bit of a mess due to the provisional commits so I thought I'd better do that stack
[14:14:15] <mvita> I recall that we (release-team) decided not to fix the -settime brokeness in 1.6.x - hence the one commit marked "WONTFIX"
[14:14:57] <wiesand> yes, that's my recollection too
[14:14:58] <mvita> (that was a few months ago - we'd have to check the archives to see details)
[14:15:18] <mvita> I suppose we could change our minds - but now we have 3 bugs -
[14:15:33] <mvita> #1 Andrew's settime deltas
[14:15:56] <mvita> #2 the out of bounds thing you and I saw
[14:16:06] <mvita> #3 the bug Mike found in review
[14:16:23] <mvita> only #1 has a valid patch
[14:16:36] <mvita> #2 has my quick and dirty WONTFIX patch
[14:16:41] <mvita> nothing for #2
[14:16:47] <mvita> sorry , #3
[14:16:51] <meffie> can someone point out the #2 item? cheyenne and i could not find it.
[14:17:05] <mvita> I updated the gerrit
[14:17:34] <mvita> 13401
[14:17:39] <meffie> ah in gerrit 11706? thanks!
[14:18:13] <mvita> yes, explained in 11706, but the patch has been 13401 all along
[14:18:52] <meffie> ah i did not make that connection. sorry.
[14:19:23] <mvita> np
[14:19:34] <mvita> it's a maze of twisty little passages
[14:19:53] <meffie> i was looking at 'deltas'.
[14:20:23] <mvita> and you found bug #3!
[14:20:48] <meffie> there are plenty more, no doubt.
[14:21:00] <mvita> heh, you said it brother
[14:21:41] <mvita> so I suggest we fix what we found and then let the thing alone
[14:21:48] <meffie> yes
[14:21:54] <mvita> nobody's using it anyway
[14:22:02] <mvita> and we can't possibly make it worse
[14:22:11] <wiesand> but for 1.6.24, we'll go with just 13403 and 13420 ?
[14:22:47] <meffie> i would take 11706
[14:22:55] <mvita> that's your call, but yes that would suffice
[14:23:08] <mvita> I agree I would like 11706 as well
[14:23:29] <meffie> because that fixes overwriting memory which is never good.
[14:23:30] <mvita> I can work up an official version of 13401 if you like
[14:23:40] <mvita> but it should go through master
[14:24:12] <meffie> it should?
[14:24:14] <wiesand> that code is still present on master?
[14:24:18] <mvita> oh, what am I saying, that code is gone in master
[14:24:24] <meffie> ah, ok.
[14:24:38] <wiesand> and on 1.8 too, right?
[14:24:41] <mvita> so I can work up an official version of 13401
[14:24:51] <mvita> yes, gone in 1.8 too, praise the Lord
[14:25:08] <meffie> yes
[14:25:32] <meffie> see, new versions can be good.
[14:25:52] <mvita> YES
[14:26:06] <mvita> So, do you want me to clean up 13401?
[14:26:24] <wiesand> yes and no
[14:26:34] <mvita> lol
[14:28:10] <wiesand> I'm really not sure whether we should touch that code at all
[14:29:02] <wiesand> why fix #1 and #2, but not #3, for a deprecated feature which has been buggy since at least 1.6.0?
[14:29:09] <mvita> like I said, that's your call and I'll accept what you decide.  I would prefer to fix things we know about - especially this panic - but I don't feel too strongly about this one
[14:30:03] meffie leaves the room
[14:30:04] meffie joins the room
[14:30:04] <wiesand> ok, let's do it
[14:30:24] <wiesand> but I'd rather not mention it in the release notes or announcement
[14:31:05] <mvita> okay, fair enough
[14:32:04] <mvita> so, just so I'm sure I've got it - you want me to clean up and resubmit 13401, correct?
[14:32:24] <wiesand> yes, please do
[14:32:28] <mvita> okay
[14:32:41] <mvita> and what would you like to do about Mike's #3 bug?
[14:33:03] <mvita> Mike, did you analyze what the impact of it is?
[14:33:24] <mvita> I think it just sets the wrong time - but I'm not sure by how much, or if it matters
[14:34:58] <mvita> I just figured out the history of it, I didn't look too closely at what the code actually does
[14:35:17] <mvita> (the bad line, that is)
[14:35:34] meffie leaves the room
[14:35:35] meffie joins the room
[14:36:19] <mvita> (looking)
[14:36:30] <meffie> not sure, might just set the time incorrectly.
[14:37:04] <meffie> it's a left over line for early refactoring. the delta variable is clobbered.
[14:37:58] <wiesand> settime is a mess
[14:38:51] <mvita> yes, and end and tv_sec are residual values from when we were populating deltas[]
[14:41:33] <mvita> okay, yes, it will be correct if there was a setTimeHost
[14:42:27] <mvita> if not, it will be the time from the last server (we heard from) I think
[14:44:00] <wiesand> ok, the fixes for #1 and #2 are very simple and clearly correct
[14:44:07] <mvita> I think we should leave that one alone… our notes about it in gerrit will suffice
[14:44:16] <meffie> agreed.
[14:44:37] <wiesand> fixing #3 will change behaviour present for 15 years - let's leave it alone?
[14:44:39] <mvita> "mostly harmless" - not a panic like the other 2
[14:44:52] <mvita> wiesand - agreed
[14:45:10] <meffie> i was trying to find the crash, #3 was just looking for the crash you and stephan found
[14:45:31] <mvita> It was an excellent find, Mike
[14:45:39] <meffie> there was no other bread crumbs we could find for the crash
[14:45:52] <mvita> My eyes boggled when I saw what you found
[14:46:02] <mvita> all those deltas - wasted!
[14:47:14] <wiesand> so, for 1.6.24pre1 we merge 11706, 13401, 13403 and 13420, update 13330, and that's it?
[14:48:46] <mvita> do we have time to squeeze in a backport of 13337?
[14:49:21] <mvita> (and 13336 prereq)
[14:50:16] <mvita> I will provide the backports, regardless
[14:50:55] <meffie> those are for 1.8.3 only, right?
[14:51:09] <mvita> oh, you may be right….
[14:51:18] <meffie> oh no, those are for 1.6.x
[14:51:35] <meffie> (mis remembered sorry)
[14:51:56] <meffie> this site that reported that bug was running 1.6.x
[14:52:54] <mvita> oh, okay
[14:53:20] <wiesand> I'd rather not block on this, but if it's ready (including sufficient review), fine
[14:53:24] <mvita> we've been having a number of fixes that are 1.8.x only or 1.6.x only, it's hard to keep them straight
[14:54:30] <wiesand> on to 1.8?
[14:54:59] <mvita> well, could you put out pre1 as you outlined and then add these for 1.6.24  final?  or does that defeat the purpose of a pre release?
[14:55:25] <wiesand> it does
[14:55:46] <mvita> okay, it can wait
[14:55:52] <mvita> the problem is quite rare
[14:56:09] <wiesand> but I'm still very busy and won't get out the pre1s all that swiftly ...
[14:56:36] <mvita> I'll do the backports, regardless, but if they don't get into pre1 they can just go in the next release
[14:56:43] <wiesand> deal
[14:56:50] <wiesand> or a pre2
[14:56:59] <mvita> deal
[14:57:45] <wiesand> good, on to 1.8 then
[14:58:00] <wiesand> 13421 seems obvious
[14:58:09] <wiesand> so does 13417
[14:58:28] <wiesand> the open question is about 13416 vs. 13440
[14:59:09] <wiesand> er, 13340
[15:00:41] <wiesand> mark, you own 13340, so we wanted to have you in the loop here
[15:00:57] <mvita> yes, thanks
[15:01:04] <mvita> and sorry for the bad patch
[15:01:12] <mvita> it's good that Andrew caught the problem early
[15:02:13] <wiesand> so we abandon 13340 and go with 13416 instead?
[15:03:17] <mvita> (looking)
[15:04:55] <mvita> sorry, I'm not seeing yet how 13416 is an alternative to 13340
[15:07:28] <wiesand> Andrew suggested it is in his comment on 13340
[15:08:05] <mvita> bear with me, I need to go get another screen to follow all this
[15:08:26] <meffie> I think stephan meant 13415 ?
[15:09:57] <wiesand> no, that one just reverts master's version of 13340
[15:10:00] <meffie> oh, 13415 is the revert on master.
[15:11:56] mvita joins the room
[15:12:16] <mvita> okay, I'm on my main computer and dual monitor now
[15:12:27] <mvita> sorry for the delay
[15:12:31] <kaduk@jabber.openafs.org/barnowl> Whoops, overslept.  Sorry for being late!
[15:12:33] <mvita> I've been ill today
[15:12:43] <meffie> d7af27f0f8 (gerrit/changes/90/13390/3, gerrit/13390/3) rx: remove rx_atomic bitops
3effb95253 (gerrit/changes/14/13414/3) rx: Statically check rx_statisticsAtomic size
b425130150 (gerrit/changes/15/13415/2) Revert "rx: fix rx_atomic warnings under Solaris"
[15:14:17] <mvita> yes, meffie, I think that stack is what we want
[15:14:53] <mvita> and I think we usually try to keep 1.6.x the same as 1.8 if possible
[15:15:19] <meffie> at any rate, we should abandon 13340. sadly.
[15:15:39] <mvita> since my bad atomics thing was already merged on master and (1.8.x?), then we would need to merge 13340 on 1.6.x just so we can revert it
[15:15:50] <mvita> or is that just silly?
[15:16:05] <mvita> I have no problem abandoning it
[15:16:39] <kaduk@jabber.openafs.org/barnowl> It seems a little silly but not excessively so
[15:18:36] mvita leaves the room
[15:19:29] <wiesand> 13340 wasn't merged on 1.8.x
[15:20:14] <meffie> also it was my bad atomics thing :)
[15:20:55] <mvita> ah, sorry, right, we've switched to 1.8.x arght
[15:21:08] <mvita> my brain is full of goo today, sorry
[15:21:22] <mvita> (sinus infection)
[15:21:39] <mvita> so yes, I will abandon it
[15:21:54] <mvita> I don't think we care if master matches 1.8.x so much
[15:22:27] <meffie> there's a revert for master, so they will match.
[15:22:27] <kaduk@jabber.openafs.org/barnowl> To some extent the whole point of having different branches is letting
them diverge, yes :)
[15:22:42] <meffie> heh, yes.
[15:23:39] <wiesand> but only to the extent really needed
[15:24:43] <wiesand> but I think we don't have to merge a known-bad change on the stable branches just to revert it
[15:24:44] <mvita> done
[15:25:31] <meffie> oh mark meant the commit history. yeah i would have never thought we would want to commit something just to have a revert commit !!
[15:26:49] <wiesand> good
[15:27:46] <wiesand> that leaves us with the question whether 12761/13416 is a valid replacement
[15:28:11] <wiesand> if I got Andrew right, it makes those rx_atomics unused?
[15:28:33] <meffie> that's my understanding
[15:28:40] <kaduk@jabber.openafs.org/barnowl> mine as well
[15:33:12] <wiesand> Mark?
[15:33:20] <mvita> I think so.
[15:33:25] <mvita> sorry, I'm fading fast
[15:34:39] <mvita> remember that even though 13340 said it was for warnings, it actually fixed a crash on Solaris SPARC caused by endianness issues
[15:35:12] <mvita> I'm looking for my notes to see if these new commit actually address that
[15:36:31] <meffie> good point, assuming the bit atomics are gone, that would obviate the endian issues too.
[15:38:27] <wiesand> 13390 still needs to be fixed
[15:39:31] <meffie> oh, yes. what in tarnation.
[15:40:00] <meffie> i'll take a look.
[15:40:10] <mvita> heh
[15:40:20] <wiesand> so, my understanding is that we don't need pullups of 13414/13415 either
[15:40:40] <wiesand> backporting a fixed 13390 would make sense though
[15:41:56] <wiesand> but it's not so urgent, since 13416 will fix the issue (or rather reduce it to warnings while building unused code)
[15:42:33] <wiesand> is that correct?
[15:44:56] <wiesand> so the plan for 1.8.3pre1 would be  "merge 13417, 13417, 13421, update NEWS, do the version strings change"?
[15:47:48] <wiesand> 13416, 13417, 13421, silly
[15:48:14] <kaduk@jabber.openafs.org/barnowl> "the keys are right next to each other"
[15:49:58] <meffie> looking..
[15:50:01] <wiesand> since there seem to be no objections or comments, on to master?
[15:50:44] <mvita> ok
[15:51:00] <meffie> https://wiki.openafs.org/devel/GerritsForStable/
[15:51:08] <meffie> looking at 1.8.x
[15:51:23] <kaduk@jabber.openafs.org/barnowl> Once more I must apologize for my inactivity on master.  I am finally
back at home though, so hopefully things will be less crazy.
[15:51:43] <meffie> stephan, if you abandon 13122    lwp: Fix possible memory leak from scan-build    then that table would be up to date
[15:54:02] <wiesand> 13122 is a reminder that there might still be a problem on master
[15:54:23] <meffie> ah ok.
[15:54:48] <meffie> (the fix for master is "kill lwp" tho:)
[15:55:41] <wiesand> (which would get us into a similar state as regarding settime)
[15:56:43] <meffie> yes, these are the ones to be merged then:
13421    vos: restore status information to 'vos status'
13417    afs: Reword "cache is full" messages    1.8.3
13416    rx: Convert rxinit_status to rx_IsRunning()
[15:57:15] <wiesand> thanks - seems we have a plan then
[15:58:56] <wiesand> anything else to discuss today?
[15:59:13] <meffie> none from me. get well soon Mark.
[15:59:18] <mvita> tx
[15:59:26] <wiesand> get well sonn Mark & Mike
[15:59:42] <kaduk@jabber.openafs.org/barnowl> get well soon, yes
[15:59:59] <wiesand> Let's adjourn then. Thanks a lot everybody!
[16:00:23] wiesand leaves the room
[16:00:25] <mvita> bye
[16:00:47] <kaduk@jabber.openafs.org/barnowl> thanks!
[16:04:00] meffie leaves the room
[17:06:14] mvita leaves the room
[17:22:32] meffie joins the room
[18:17:38] mvita joins the room
[19:20:04] mvita leaves the room
[20:50:53] meffie leaves the room