- Thursday, October 14, 2021

release-team@conference.openafs.org

Thursday, October 14, 2021< ^ >

kaduk@jabber.openafs.org/barnowl has set the subject to: openafs release team

Room Configuration

Room Occupants

GMT+0
[11:55:46] mbarbosa joins the room
[13:51:23] Cheyenne joins the room
[14:48:52] meffie joins the room
[15:58:43] wiesand joins the room
[16:00:22] <kaduk@jabber.openafs.org/barnowl> greetings
[16:00:57] <meffie> is it thursday already?!
[16:01:49] <Cheyenne> yes it is..
[16:01:54] <Cheyenne> Hello all
[16:02:13] <kaduk@jabber.openafs.org/barnowl> It's hard to believe, but yes, it's thursday
[16:03:47] <wiesand> Hello
[16:05:01] <kaduk@jabber.openafs.org/barnowl> Hi wiesand!
[16:05:08] <wiesand> I worked on bookkeeping regarding the outstanding FBSD changes
[16:05:30] <meffie> excellent! thank you.
[16:05:46] <wiesand> I also pulled up 14830 (Linux 5.14). There was a path conflict with 14811, so I merged that one before.
[16:06:42] <kaduk@jabber.openafs.org/barnowl> makes sense (And good to get the preprocessor indentation consistent
across branches)
[16:06:47] <wiesand> Pulling up 14791 (Linux 5.15) isn't quite that trivial.
[16:07:16] <Cheyenne> I have a 1.8.x version of the linux 5.15 patch if needed.
[16:07:57] <kaduk@jabber.openafs.org/barnowl> might be unavoidable, given the madness that 14791 had to remove
[16:10:46] <wiesand> There's a conflict with (at least) 14825, and thus earlier changes like 13602 (rx: Remove unneeded rxi_ReceiveDataPacket params). And probably more - that's how far I got before the meeting.
[16:11:30] <wiesand> I haven't made up my mind how to resolve this yet. I hate *adding* to the code skew.
[16:12:23] <Cheyenne> the conflict with 13602 is just a context change, no code changes needed
[16:12:26] <wiesand> Proposals welcome.
[16:12:45] <kaduk@jabber.openafs.org/barnowl> This is changing the API for how we print things.  So if we have any
differences in what we ever print across branches, there is going to
be skew/conflicts for this change.
[16:13:31] <Cheyenne> I can push my pick to github and you can look at it from there if you wish.
[16:13:46] <wiesand> The clash in rx_internal.h is trivial to resolve. There are two more files though I haven't looked at yet.
[16:14:53] <wiesand> Cheyenne: you can just as well submit it to gerrit, if you don't mind that I *may* ask you to abandon it
[16:15:18] <Cheyenne> that's fine, or you can just take it over. Either way is fine
[16:18:24] <wiesand> As I said, I'd like to have a closer and more complete look. Adding to the code skew is bad, pulling something like 13602 into a point release to avoid that isn't pretty either.
[16:19:30] <wiesand> That's what I have to report today. I'm hoping for a 1.8.8.1 next week.
[16:19:32] <kaduk@jabber.openafs.org/barnowl> Is 13602 really so bad?  The commit message says it's just an internal
refactoring.
[16:20:29] <Cheyenne> the conflict is simply because the change for 5.15 added a line at the bottom of the .h file
[16:20:48] <Cheyenne> anyway -- https://gerrit.openafs.org/13602
[16:20:50] <Cheyenne> woops
[16:21:39] <Cheyenne> https://gerrit.openafs.org/#/c/14831/1
[16:22:30] <Cheyenne> and the conflict with 14825 is a change in context.
[16:23:32] <meffie> in context?
[16:24:04] <kaduk@jabber.openafs.org/barnowl> The lines around the lines that actually change
[16:24:49] <kaduk@jabber.openafs.org/barnowl> aka the "context" in the "diff context" output, and even what you get
for a "context diff" (but who uses "context diff" rather than "unified
diff" anyway?)
[16:24:50] <meffie> oh, sure but that is literally every merge conflict ;)
[16:25:14] <kaduk@jabber.openafs.org/barnowl> Soemtimes there's a conflict in the "old" lines that are being
changed, too!
[16:25:57] <wiesand> Again, I'd like to look this in more detail, and won't manage during this meeting. It may well be possible that it can be done by adding a few clean cherry picks of harmelss other changes.
[16:26:01] <Cheyenne> anyway I was able to simply rebase the 5.15 change on top of 14825 (didn't with no conflicts)
[16:26:17] <meffie> yes. i guess we couldn't have done a fix that retains the old (osi_Msg...) on master for backporting
[16:27:23] <meffie> we'd still have conflicts because we still would fix them to be osi_Msg()..
[16:28:38] <Cheyenne> there aren't that may calls to osi_Msg.  (7 calls across 2 files)
[16:28:39] <wiesand> I didn't propose to retain (osi_Msg...)
[16:28:40] <meffie> thanks Cheyenne
[16:31:30] <Cheyenne> On a different note... autoconf-2.70 is generating some warning messages about some obsolete macro usage: AC_CONFIG_HEADER, AC_PROG_LEX, AC_HEADER_STDC and AC_HEADER_TIME
[16:32:33] <Cheyenne> It doesn't appear to have broken anything, just some warning messages when running configure.  (my system recently got an update for autoconf to 2.71).
[16:33:07] <meffie> we update the macros from time to time. ;)
[16:33:22] <kaduk@jabber.openafs.org/barnowl> Yeah, it should be fine to tweak configure.ac to avoid those new
warnings
[16:34:03] <Cheyenne> I looked at autoconf-2.64 and those macros are marked as obsolete even then.  the change with 2.70 is that autoconf is now giving warnings on them
[16:34:18] <meffie> ok, thanks.
[16:34:59] <kaduk@jabber.openafs.org/barnowl> At least a couple look familiar from other projects that moved off
them, yes
[16:35:26] <wiesand> Just mind the path conflict with those "version string" changes (like 14828)
[16:35:39] <Cheyenne> the only one that I have a little grief with is AC_HEADER_TIME and a check in roken.h.in that uses it.
[16:36:06] <Cheyenne> all the others appear to be easy to remove (or change for the AC_PROG_LEX)
[16:37:48] <kaduk@jabber.openafs.org/barnowl> Probably worth checking if upstream heimdal has adjusted their
roken.h.in usage of AC_HEADER_TIME -- our snapshot is years old
[16:38:08] <Cheyenne> did that.. it hasn't changed.
[16:38:18] <kaduk@jabber.openafs.org/barnowl> Ah, okay.
[16:38:29] <kaduk@jabber.openafs.org/barnowl> You could file a bug on github if you want ;)
[16:38:29] <Cheyenne> I'm thinking of simply doing a define for TIME_WITH_SYS_TIME for that one case
[16:42:20] <Cheyenne> anyway -- clean master build against linux-5.15-rc5+, and adding the 5.15 patch to 1.8.x I get a clean build there as well.
[16:42:38] <meffie> excellent, thanks!
[16:42:40] <kaduk@jabber.openafs.org/barnowl> happy to hear it
[16:43:09] <kaduk@jabber.openafs.org/barnowl> I pulled 14826 into debian already, for the improved 5.14 support
[16:44:24] <Cheyenne> I let the gentoo folks know as well
[16:46:08] <wiesand> The community was pretty helpful with this one. We probably should mention that in the 1.8.8.1 announcement.
[16:46:13] <kaduk@jabber.openafs.org/barnowl> +1
[16:47:22] <meffie> +1
[16:47:53] <Cheyenne> Yes.. finding the linux commit that caused the error was helpful.  I forgot to mention Mark Vitale for giving me a hint on how to possibly recreate the problem
[16:49:30] <wiesand> So, thanks Mark & Cheyenne!
[16:49:42] <meffie> hear hear
[16:49:45] <wiesand> Other topics for today?
[16:49:55] <meffie> i have nothing
[16:50:11] <Cheyenne> nothing more from here..
[16:50:16] <kaduk@jabber.openafs.org/barnowl> not really.  I merged a couple things, but boring prep-work type stuff
[16:50:30] <meffie> thank you!
[16:51:06] <Cheyenne> wiesand: if you need help with the 5.15 patch, let me know. You can mark my commit as abandoned or just change the commit yourself.
[16:51:13] <meffie> marcio an mark did reviews on the bozo-large-string topic.
[16:51:16] <kaduk@jabber.openafs.org/barnowl> I left a question for Andrew on 14802 but am sure he will see it in
due time
[16:51:33] <kaduk@jabber.openafs.org/barnowl> thank you for the reviews on bozo-large-strings!  I am slowly working
my way through them as well.
[16:51:43] <meffie> excellent, thank you!
[16:52:38] <kaduk@jabber.openafs.org/barnowl> Sounds like time for a motion to adjourn
[16:52:47] <meffie> one strange question. at what point are we going to drop support for the solaris 32bit "lack of filedescriptors"  stuff?
[16:52:52] <Cheyenne> wiesand: there is another cleanup commit that was merged recently: 14629 (only indentation and whitespace changes in rx.h)
[16:52:57] <wiesand> Motion sustained
[16:53:10] <meffie> :)
[16:53:25] <kaduk@jabber.openafs.org/barnowl> meffie: I have no idea.  Both on the timeline and on what the thing is
:)
[16:53:28] <wiesand> Thanks for the hint. I'll have a look at 14629.
[16:53:29] <meffie> :)
[16:53:35] <Cheyenne> everyone have a good week
[16:53:52] <kaduk@jabber.openafs.org/barnowl> yes, thanks all, and have a good week
[16:54:09] <meffie> ok have a good week all!  (i'll move that weird question to gerrit ;)
[16:54:22] <meffie> thanks
[16:54:27] <wiesand> Thanks a lot everybody!
[16:54:58] meffie leaves the room
[16:55:48] wiesand leaves the room
[19:52:33] meffie joins the room
[19:52:52] meffie leaves the room
[21:19:32] mbarbosa leaves the room
[23:32:18] Cheyenne leaves the room