- Thursday, September 16, 2021

release-team@conference.openafs.org

Thursday, September 16, 2021< ^ >

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

Room Configuration

Room Occupants

GMT+0
[11:59:34] mbarbosa joins the room
[12:18:25] mbarbosa leaves the room
[12:18:37] mbarbosa joins the room
[13:43:31] Cheyenne joins the room
[14:46:16] meffie joins the room
[15:55:58] <kaduk@jabber.openafs.org/barnowl> Looks like I will be a few minutes late.  I should be able to quickly
re-review and merge the latest revision of the linux 5.15-support
patch when I get back.
[15:56:33] <Cheyenne> Good morning everyone
[15:56:52] <Cheyenne> (or afternoon|evening)
[15:57:43] <meffie> good day
[16:05:13] wiesand joins the room
[16:06:02] <wiesand> Good evening. Sorry, it took several minutes to get in today.
[16:06:43] <meffie> welcome!
[16:08:14] <wiesand> Not much to report from my side today. I plowed my way through the fbsd12-build and dbsd12-ro stacks, and merging those is next on my list.
[16:09:12] <wiesand> But it looks like FBSD is more like a can of worms than I anticipated.
[16:09:24] <meffie> oh, sorry.
[16:09:38] <meffie> because some common code changed?
[16:11:20] <wiesand> No, it's mostly clearly FBSD specific, though sometimes in a piece of common code with an "#if defined FBSD_*" clause.
[16:14:17] <wiesand> But after merging those, Buildbot will hopefully help.
[16:16:03] <wiesand> But quite a few FBSD changes seem to have been merged on master lately, and quite a few more are still open
[16:16:12] <kaduk@jabber.openafs.org/barnowl> Indeed
[16:16:29] <kaduk@jabber.openafs.org/barnowl> I tried to get through most of that stack, last week.
[16:16:48] <kaduk@jabber.openafs.org/barnowl> There were a few that I commented on and waited for a response, but
probably don't actually need any revisions before merging.
[16:18:12] <wiesand> I noticed commits touching "afsweb" stuff. Maybe that should rather be removed?
[16:18:32] <kaduk@jabber.openafs.org/barnowl> probably, yes
[16:18:32] <meffie> cheyenne and i were just talking about that :)
[16:19:13] <meffie> we were going to wait until kdump deorbit hits
[16:19:17] <Cheyenne> I just marked two of the static-analysis commits as abandoned since it probably makes more sense to just remove the associated code (afsweb and jafs)
[16:19:18] <wiesand> And what is "libjafs" anyway?
[16:19:27] <meffie> src/JAVA
[16:19:55] <meffie> for afsweb if i recall
[16:20:45] <Cheyenne> Also Ben, there is a doc update that cleans out a afsinstall reference in src/README.md
[16:20:54] <wiesand> "just remove the associated code" sounds reasonable…
[16:21:06] <kaduk@jabber.openafs.org/barnowl> I saw that one, thanks!
[16:21:24] <Cheyenne> thanks to meffie for spotting it.
[16:21:29] <wiesand> Once it's dumb question time, what's this UKERNEL stuff about anyway?
[16:21:33] <meffie> sorry i missed that earlier :(
[16:21:53] <meffie> libuafs, userspace cache manager (mostly for fuse)
[16:22:54] <wiesand> ok, thanks, I wouldn't want to drop the fuse client at this point
[16:22:57] <kaduk@jabber.openafs.org/barnowl> What Mike said.
I forget if afsio uses UKERNEL or operates at a lower layer
[16:23:21] <meffie> i do not think it does
[16:23:42] <kaduk@jabber.openafs.org/barnowl> That's my suspicion, yes, but I'm not sure offhand
[16:24:09] <meffie> but yes, please dont drop the fuse client :) we should add authentication to it.
[16:25:10] <Cheyenne> Linux-5.15-rc1+ report -- clean master build with patch (thanks Ben for merging it).
[16:25:12] <wiesand> and some substitute for pioctls IIRC?
[16:25:21] <meffie> woot! thanks Cheyenne
[16:25:53] <meffie> correct, some alternative to pioctls would be needed
[16:26:25] <wiesand> Thanks Cheyenne! That clearly ranks high on the 1.8.9 list (and is unlikely to clash with the FBSD changes we have in the pipeline).
[16:26:52] <Cheyenne> Yes -- it should be a fairly simple pull up
[16:27:04] <wiesand> And yay, 14779… thanks Ben!
[16:27:46] <kaduk@jabber.openafs.org/barnowl> :)
[16:28:50] <Cheyenne> BTW -- if there aren't any odd compiler problems (thinking of AIX) with the 5.15 patch, there is probably some additional code cleanup that could be done at some point.
[16:31:08] <wiesand> At this point I'm leaning towards making 1.8.9 a "lean" release. Serious FBSD improvements, Linux 5.15, a few more minor features and fixes (in particular, 14740 and 14594). Thoughts?
[16:32:13] <kaduk@jabber.openafs.org/barnowl> Okay by me
[16:32:16] <Cheyenne> Also linux kernel related, looks like they are bumping the minimum compiler version to gcc 5.1 and will be allowing --std=gnu11
[16:32:36] <kaduk@jabber.openafs.org/barnowl> Cheyenne/meffie: see comment on 14085 re afsinstall
[16:34:01] <Cheyenne> oops.. will fix (14805).
[16:34:04] <meffie> 14805 i think
[16:34:42] <kaduk@jabber.openafs.org/barnowl> oops, sorry for typo
[16:35:22] <Cheyenne> ditto
[16:35:35] <Cheyenne> I just pushed an update
[16:35:38] <meffie> hey, typos are my job!
[16:35:40] <wiesand> +1 on the missing "not" in the 14805 commit message
[16:35:41] <kaduk@jabber.openafs.org/barnowl> thanks
[16:36:46] <meffie> +1, thanks.
[16:36:58] <wiesand> hmm, "not did remove" still sounds a bit funny, but then it's not my native language thus it could be me…
[16:37:06] <Cheyenne> Also looks like the upstream linux idea of using -Werror in the linux kernel builds has been changed to using it just for test builds
[16:37:44] <Cheyenne> wiesand: English is my native language and I still mess it up :)
[16:38:21] <wiesand> re Werror: Sounds like weakening the standards - what are the consequences for us?
[16:38:26] <kaduk@jabber.openafs.org/barnowl> It does sound a bit funny, but I think it is a valid construction.
[16:39:18] <Cheyenne> It's actually the other way around.  They were going to turn on -Werror for all kernel builds but were running into push back from the kernel community
[16:39:26] <Cheyenne> https://www.theregister.com/2021/09/08/compromise_linux_kernel_compiler_warnings/
[16:39:37] <kaduk@jabber.openafs.org/barnowl> Wow, it made The Register even
[16:41:28] <wiesand> kind of sad IMO...
[16:45:17] <meffie> ok, do we have anything else?
[16:45:51] <wiesand> I remember we had (F77) physics code running on all kinds of platforms (IRIX, AIX, HPUC, SunOS) w/o compiler errors. When we introduced Linux (with the venerable g2c) it was the most picky platform to compile on and required quite a few changes. And it was a huge win.
[16:45:54] <Cheyenne> I don't have anything.  wiesand, if you need any help with the linux-5.15 patch let me know.
[16:45:55] <kaduk@jabber.openafs.org/barnowl> I don't think I'm waiting on other people for much of anything ...
just need to review review review
[16:46:10] <meffie> ok, thanks!
[16:46:21] <wiesand> I don't have anything else today. Let's adjourb.
[16:46:21] <kaduk@jabber.openafs.org/barnowl> I'm so sorry ... F90 is so much nicer than F77 ;)
[16:46:36] <kaduk@jabber.openafs.org/barnowl> Sounds good.
Thanks everyone!
[16:47:18] <wiesand> The IRIX compiler would even allow you to overwrite the constant "4" with "5" but hide the problem from you…
[16:47:33] <wiesand> Thanks a lot everyone!
[16:47:39] <meffie> (F77, that's the last time i used fortran, back in 1987)
[16:47:53] <meffie> have a good weekend!
[16:48:04] <Cheyenne> Have a good week everyone.
[16:48:14] <wiesand> Cheyenne: Thanks, will let you know if I need help!
[16:48:27] meffie leaves the room
[17:25:10] wiesand leaves the room
[17:33:50] meffie joins the room
[17:34:53] meffie leaves the room
[21:50:14] mbarbosa leaves the room
[22:33:52] Cheyenne leaves the room