[04:23:41] --- Simon Wilkinson has become available [04:24:32] IRIX should fall back to using mutexes. Derrick has a commit to fix that. [04:24:42] I've just pushed the moderator queue for openafs-devel [07:04:16] > I've just pushed the moderator queue for openafs-devel some of those were dups alas. [07:20:01] --- mdionne has become available [07:55:30] Yeah, well, some of them were so old I had no way of knowing. [07:55:42] We really need to push the queue more than once every 6 months. [08:34:34] while true re 6 months, there is absolutely a way of knowing: looking at them [08:35:11] Not unless you have a 6 month knowledge of what's gone past on the lists. [08:35:29] hi. [08:35:53] Yeah well, you want to flush the mail queues on a regular basis, be my guest. [08:36:21] honestly, i forgot [08:36:37] We need to do something about our mailing lists. [08:36:45] I'd be highly tempted to just give them all to Google. [08:36:56] the problem is the lack of moderator. giving them to google won't fix that [08:37:20] Well, it reduces the load on the moderator, because google groups are pretty good at catching spam. [08:37:49] Either that, or we should front the lists with TDMA. In the absence of a moderator with time, the aim should be to remove the need for a moderator. [08:38:05] time division multiplexing algorithm? [08:38:26] or we could simply refuse to accept mail that is from non-subscribers [08:38:34] yes, that would have been my suggestion [08:38:36] --- ksumner has left [08:38:44] --- meffie has left [08:38:45] Sorry, TMDA. [08:39:08] I don't think we can refuse to accept mail from non-subscribers, whilst we list openafs-info with a mailto: link on the OpenAFS web site. [08:39:35] --- ksumner has become available [08:39:37] well, that's also editable :) [08:39:42] And I do think that there are arguments for accepting non-subscriber posts. I would just like a way of accepting them which doesn't take anyones time. [08:39:49] i'd like a pony [08:40:07] also for mysql remote to actually support the same syntax as "i'm running the cli on the server" [08:40:27] actually, given mysql syntax i'd forgo the pony [08:43:21] ugh. the original submission for 2890 built fine on irix but when merged to the current master it fails. [08:45:17] That probably means that current master is broken on Irix, rather than being a problem with just that patch. [08:45:28] > moderators How many people have moderation access? [08:45:53] well, sort of. the original stats split patch doesn't include enough headers to build in the kernel [08:46:01] i'll have an answer in a moment [08:46:15] Irix is missing the definition of rx_atomic_mutex [08:46:52] Or rather, the definition of the afs_kmutex_t structure, I suspect. [08:46:52] no, its missing the definitioin of kmutex_t [08:48:19] what I'm not understanding is why http://buildbot-master.secure-endpoints.com:8010/builders/irix-builder/builds/477 succeeded. [08:48:44] --- meffie has become available [08:48:56] because it was against the old master that didn't bother building rx_stats in the kernel perhaps? i didn;t look yet [08:48:59] Patchset 3 is wrong [08:49:35] Or rather, it needs to be built on top of Marc's change that protects string.h from the kernel. [08:50:09] Derrick can you update your branch and resubmit? [08:55:06] I canceled the 2890/3 buildbot build so 2890/4 can proceed [08:56:13] Well, I guess we'll know in 2 hours or so :( [08:56:48] Derrick: Do you know whether we can safely use xlc_r to compile the LWP code on AIX. If we could, we could make the build system a bit simpler ... [08:56:56] the last successful irix build took 1h 1m 54s [08:57:05] >before, the symbols were referenced but missing, so the module just didn't load Ideally we'd catch this at build time. looks tricky, because modpost insists on using -w (warn instead of error for missing symbols) for external modules [08:57:18] not a clue [08:57:44] marc, yes, i was wondering about doing that [08:57:45] it would be nice if someone with an aix system setup a build bot slave [08:58:05] It would be nice to get HPUX, AIX and Solaris going. [08:58:06] the aix system i have isn't mine. i can ask [08:58:24] I guess we could do Solaris x86 pretty easily, at least with one of the final OpenSolaris seeds. [08:58:24] i have a solaris vm. i sshould find time. not this week [08:58:49] i have solaris 9 and 10 vms. in fact i think thoughtpolice.co.uk has "starter" vms for it [09:40:21] --- jaltman has left: Replaced by new connection [09:40:22] --- jaltman has become available [10:06:48] --- jaltman has left: Disconnected [10:06:58] --- jaltman has become available [10:44:14] --- jaltman has left: Replaced by new connection [10:44:15] --- jaltman has become available [10:49:47] --- jaltman has left: Replaced by new connection [10:49:48] --- jaltman has become available [10:52:09] --- jaltman has left: Replaced by new connection [10:52:10] --- jaltman has become available [10:54:55] the irix build succeeded. I pushed the patch. please report any additional build issues with master. [11:13:42] --- jaltman has left: Disconnected [11:13:54] --- jaltman has become available [11:29:56] I'm looking at the things we could do to make the Unix build system nicer. I think there's going to be a very fine line between clever and voodoo. [11:31:20] Having panic'd my machine from using a clang-compiled module, I'm really tempted to look at using bsd.kmod.mk right now ... but I have to give a presentation tomorrow. [11:33:54] I've been using clang to build our userspace, primarily for the static analysis bits. Not sure I'd trust it with a kernel module ... :) [11:35:10] --- Russ has become available [11:35:49] People are apparently running entire kernels built with it (but presumably it needs slightly different CFLAGS than gcc). [11:36:29] Yeah. We get that vaguely right in userspace, but kernel space is a whole different ball game. [11:36:58] Right. "But if I use bsd.kmod.mk, someone else has already worried about that for me." [11:37:19] Yeah. That's much the same as we do for Linux. [11:37:46] There's a perl script that transforms the the Makefile.common into a Makefile that can be fed into the Linux build system, and then we just run as part of that system. [13:21:09] --- deason has become available [14:48:27] --- mdionne has left [17:55:46] --- jaltman has left: Replaced by new connection [17:55:48] --- jaltman has become available [18:29:50] --- jaltman has left: Disconnected [18:30:03] --- jaltman has become available [19:45:13] --- jaltman has left: Replaced by new connection [19:45:14] --- jaltman has become available [20:13:46] --- jaltman has left: Disconnected [20:13:52] --- jaltman has become available [20:53:54] --- jaltman has left: Disconnected [20:54:19] --- jaltman has become available [22:24:00] --- deason has left [23:22:04] --- Russ has left: Disconnected