- Wednesday, May 20, 2015

release-team@conference.openafs.org

Wednesday, May 20, 2015< ^ >

Room Configuration

Room Occupants

GMT+0
[13:25:15] meffie joins the room
[13:49:56] wiesand joins the room
[14:00:33] kaduk joins the room
[14:00:51] Jeffrey Altman joins the room
[14:00:54] <wiesand> Hi
[14:01:01] <kaduk> $timeofday
[14:02:05] <wiesand> This is likely to be a brief meeting I think.
[14:02:26] <meffie> ok
[14:02:57] <wiesand> Linux: no further surprises.
[14:03:46] <wiesand> linux-4.1: I pulled those up. They need review and verification.
[14:04:11] <wiesand> Is any extension to LINUX24 needed there?
[14:04:34] <kaduk> I doubt it; linux24 is inherently not compatible with linux 4 == (linux26)
[14:05:11] <wiesand> Good.
[14:05:56] <wiesand> 11124 (mmap) needs some more review, but let’s hope it’s the final one.
[14:06:28] <wiesand> And then there’s 11870, to go on top of yosemite/clang (or not). Opinions?
[14:06:30] <kaduk> (I just pushed a new 6947, Jeff's previous claims about a new version this morning aside)
[14:07:07] <kaduk> I do not particularly care for 11870's approach; I +1'd it because it solves the actual problem I found.
[14:07:30] <kaduk> Since it's a file-local helper, Jeff's asprintf()+free() suggestion is probably best
[14:07:54] <kaduk> (If it was a larger scope, we might have to think about whether the burden of the API contract involving the caller free()ing output arguments is worth it.)
[14:08:19] <wiesand> Saves some stack space too?
[14:08:34] <kaduk> yeah
[14:09:47] Jeffrey Altman leaves the room
[14:09:54] <wiesand> I won’t get around to that anytime soon... Any volunteers?
[14:10:00] <kaduk> Hmm, though there is some awkwardness about the caller wanting to avoid handling one of the outputs.
[14:10:14] <kaduk> I will see what I can get done during the meeting.
[14:10:37] <kaduk> Actually, no, it's not awkward, since the char* become char**
[14:10:59] <wiesand> That’s the last thing on my list for 1.6.12. Anything missing?
[14:11:24] <meffie> i have nothing new.
[14:11:51] <kaduk> Mike: good, go look at softsig ;)
[14:12:13] <wiesand> So, with a little luck, pre2 “soon”.
[14:12:43] <meffie> heh, ok! :)
[14:13:01] <wiesand> So, on to 1.8?
[14:14:06] <kaduk> We got some good comments on 6947, and have a new version up.
[14:14:49] <kaduk> Did we talk about the bosserver SIGFPE avoidance options last week? I forget
[14:15:01] <meffie> yes
[14:15:40] <meffie> i've not pushed the alternative patch for it, tho, i think i like the first one better anyway.
[14:16:19] <wiesand> SIGFPE was even mentioned in the minutes ;-)
[14:16:32] <kaduk> So much for my memory :(
[14:16:54] <wiesand> I had to look it up, and I wrote them ;-)
[14:17:17] <kaduk> Anyway, I was thinking about it some yesterday, and regardless of what we do, it will be a breaking behavior change (i.e., POLA violation) that needs to get mentioned in the release notes. So, we have a lot of freedom to do whatever we feel is best.
[14:17:52] <meffie> it's not something that is done very often, really.
[14:18:33] <kaduk> Yup.
[14:19:09] <meffie> also, i find the -mode 0 flag strange and confusion.
[14:19:21] <wiesand> With my admin hat on, I’m fine with “touch, then kill”.
[14:19:56] <meffie> ok, thank you.
[14:21:53] <wiesand> Er, “touch, then bos srestrict”
[14:22:37] <wiesand> Anything else to discuss today?
[14:23:34] <kaduk> Hmm, just what we want to do next for 1.8 if softsig goes in, I guess.
[14:23:58] <kaduk> Perhaps just the bosserver/FPE and softsig conversions will keep us busy.
[14:24:04] <meffie> i have nothing. btw, the rx_address stuff is intented for post 1.8.x, if that wasnt obvious.
[14:24:10] <kaduk> :)
[14:24:32] <kaduk> I'm just waiting for someone to notice the commit mail and ask when ipv6 support will be ready.
[14:24:42] <meffie> there's a ton of patches on top of that in github, but not diced for gerrit yet.
[14:24:52] <kaduk> *nods*
[14:25:46] <wiesand> But it’s good to see activity there.
[14:26:29] <wiesand> NB why was softsig needed after all?
[14:27:23] <kaduk> a chain of hairy yaks
[14:27:25] <meffie> so, we can free up SIGUSR{1,2}
[14:27:57] <kaduk> we want log rotation at the major version boundary, but log rotation normally uses SIGUSR1 which is already in use, so we need new softsig to free it up
[14:28:00] <meffie> because then we can use those for new things, like for the external log rotation.
[14:28:33] <meffie> SIGHUP is already used for reseting tracing
[14:28:36] <wiesand> So, maybe that’s the logical candidate for the change to pursue next?
[14:28:56] <kaduk> probably
[14:29:49] <meffie> > chain of hairy yaks
[14:30:00] <wiesand> http://gerrit.openafs.org/#q,status:open+project:openafs+branch:master+topic:externalize-log-rotation,n,z ?
[14:30:03] <meffie> ha, yes, yak shaving :)
[14:30:30] <kaduk> yeah, that one
[14:31:04] <wiesand> Should keep folks busy til next week’s meeting ;-)
[14:31:19] <kaduk> deorbiting linux24 was part of the same yak chain as well, since the new softsig is incompatible with linux24's LinuxThreads
[14:31:37] <meffie> yes, thank you for that kaduk!
[14:32:39] <meffie> (as is the SIGFPE stuff, same yaks:)
[14:33:16] <wiesand> Fine. If there’s nothing further to discuss, I could go attach inventory labels to some new hardware...
[14:33:29] <kaduk> Ouch.
[14:33:37] <kaduk> I'm sure we can come up with something to discuss if we try hard enough...
[14:33:46] <meffie> lol
[14:34:27] <wiesand> Thanks, but it’s not that bad.
[14:35:34] <wiesand> Ok, I’ll run. MInutes should go out later. Unless my provider kicks me out like a fortnight ago...
[14:35:58] <meffie> good day. thank you wiesand.
[14:36:03] <wiesand> Thanks a lot, and Goodbye.
[14:36:07] wiesand leaves the room
[14:36:08] <kaduk> thanks
[14:36:24] <meffie> have a good day dr. kaduk
[14:36:34] <kaduk> you, too
[14:37:40] meffie leaves the room
[16:12:53] shadow@gmail.com/barnowl0396B1CB leaves the room
[16:13:04] shadow@gmail.com/barnowl0396B1CB joins the room
[18:25:04] shadow@gmail.com/barnowl0396B1CB leaves the room
[18:25:12] shadow@gmail.com/barnowl0396B1CB joins the room
[21:20:18] kaduk leaves the room