[00:18:18] --- jaltman has left: Disconnected [00:39:23] --- Jeffrey Altman has become available [00:47:01] --- jaltman has become available [01:11:14] --- jaltman has left: Disconnected [01:14:08] --- Russ has left: Disconnected [01:41:09] --- haba has become available [01:41:47] --- kaj has left [01:41:51] --- kaj has become available [03:33:08] missed wakeup on freebsd: i promise i will look at the issue today [03:33:33] i need to build a kernel on my vm as the one it came with did not have its build byproducts included. [03:34:15] (and the kernel i did build for the vm, i then trashed the vm, and had to rerestore from the zip file) [03:48:20] --- Jeffrey Altman has left: Replaced by new connection [03:48:20] --- Jeffrey Altman has become available [06:45:00] jaltman - I have not yet tried the 64-bit one (have to first install a 64-bit win7) [07:18:18] --- deason has become available [07:47:33] shadow: note that the patch you pushed for me breaks the openafs build for 8.0 systems -- the ngroups_max tuneable only exists in current and RELENG_8. Quick workaround coming soon, and proper fix probably tonight (probably involving creating FBSD81_ENV). [07:47:57] We didn't actually work in 8.0, though did we? [08:01:57] we probably worked in 8.x at some point but i think recent changes had already broken it [08:13:30] Well, "work" is a slippery concept. [08:18:27] I can't see how setpag can have worked on any kernel version where syscalls took a thread rather than a proc. [08:20:39] Nor can I ... but apparently Matt claims that he had a client that works at some point in there. [08:22:19] setpag isn't really high on my list of features in a working client. but then, i have a mac :) [08:33:33] Where did this openafs-devel-1_5_73_1 thing come from? [08:33:48] uh. ask a better question [08:34:11] we plan to do point releases. i decided now would be a good time to start [08:35:05] well, the issues that motivated it were the aklog segfault, and something with fsevents, right? [08:35:46] Hmmm. It occurs to me that this point release doesn't match the way we're intending on doing them on stable, though. [08:35:59] fsevents [08:36:00] As in, it was cut from master, so it's got everything that's happened since 1.5.73. [08:36:15] Rather than _just_ fsevents - which is the way we'd do it if we did it on stable. [08:37:17] the aklog segfault only happened to people with the srv record thing. [08:37:25] --- haba has left [08:37:35] yeah. for stable i'd have branched [08:37:43] this is just "i don't want to commit jeff to building" [08:42:33] Hm, I missed discussion about the aklog segfault. I bet there's a mailing list or two that I should be on ... [08:42:46] port-darwin was where it hit [08:43:12] Mutter, port-* mailing lists. Mutter. [08:43:31] "Let's all have a mailing list each" [08:43:35] well, they were intended as "devel issues for only these people that no one else cares about" [08:44:16] and we created a solaris one recently so the solaris folks who wish to help us but don't have time for -devel can interface with us [08:44:56] Yeah - I think that one makes sense. I just worry that the other just fragment traffic so the right people don't see it. [08:45:40] that was the original worry i had when we set them up [08:45:49] but this is almost 10 years out [08:46:18] i wanted to kill port-darwin once the port was done. people on it didn't want to interact with others necessarily, and so we kept it. [08:46:31] like, mac users who wanted to be mac users and ask mac-specific questions [08:47:03] That's fair. I should just get over myself and join the list, I guess. [08:47:23] it took me quite a while to get over myself and do so [08:48:22] --- abo has left [08:48:40] --- abo has become available [08:49:27] Crap, forgot to put "Reported By" in the commit message. [08:49:38] reported by? [08:51:14] Emmanuel Vadot -- simon pointed him to me from #openafs [08:51:47] Hope that was OK! [08:52:13] (By the way, Simon, feel free to point people to kaduk@mit.edu; I'm kind of trying to deprecate the gmail address, but it has several years of easily-searchable mailing list history that I would miss if I dropped it completely.) [08:52:30] It was definitely fine -- I probably would have missed the bug if he hadn't reported it. [08:52:41] i forward all mail to gmail so i get the searching history [09:14:27] --- mho has left [09:16:56] --- mho has become available [09:26:46] --- kaj has left [09:33:26] --- reuteras has left [10:53:55] --- kaj has become available [11:07:58] --- Russ has become available [12:06:37] What is the general solution for getting a diff between two different versions of a change in gerrit? (I'm assuming that it was just the ANSI-fication this morning.) [12:07:33] you want a diff, or you want to merge? there's some interdiff feature in gerrit but i've never run across it [12:07:57] I would want a diff between the diffs. [12:08:52] i don't know. i can tell you what i changed: i fixed the k&r-prototyped function to have an ansi prototype [12:09:08] That's what it looked like, yes. [12:10:36] You can coerce Gitweb into giving you a diff between two arbitrary commits: http://git.openafs.org/?p=openafs.git;a=commitdiff;hp=changes/46/1546/1;h=changes/46/1546/2 but that doesn’t help you if the base commit changed (which it did in that example). [12:11:44] Probably the easiest way is to pull the two refs into your local repository and diff <(git show firstcommit) <(git show secondcommit). [12:12:08] i'm fairly certain gerrit can tell you, somehow. [12:12:49] Anders: so, that would be manually fetching from gerrit's HEAD:changes/foo/bar ? [12:13:07] yes. [12:13:15] (tho i'm not anders) [12:13:39] --- jaltman has become available [12:13:40] "you'll do." [12:14:09] i have the freebsd vm in front of me, the wakeup issue... i want to look at a macos issue then i will go back to it. i have an idea as to the issue [12:14:20] Yay. [12:14:45] Can you attach (k)gdb to your VM's serial console? [12:15:39] i should research that. i have no idea yet [12:16:15] apparently yes, if i bring up a second vm [12:16:59] It might be useful. I think that it can do a better job at getting complete backtraces than the compiled-in kdb can. [12:21:35] if you had asked me about building 1.5.74 I would have said to please wait until the end of the week so I can have confirmation from David regarding the afslogon issue on 64-bit. [12:22:15] which i figured, hence 1.5.73.1 :) [12:22:30] is ok, trying to fix an issue which finder wets its nest [12:44:47] --- steven.jenkins has left [12:51:32] --- jaltman has left: Disconnected [12:55:46] --- steven.jenkins has become available [12:57:38] --- steven.jenkins has left [12:59:39] --- steven.jenkins has become available [13:07:47] --- jaltman has become available [13:14:01] --- deason has left [13:16:16] --- deason has become available [13:17:24] > we created a solaris one recently so the solaris folks who wish to help us but don't have time for -devel can interface with us Was that announced somewhere and I just missed it? [13:20:02] It was discussed both here and on openafs-devel as part of a thread initiated by a member of the OpenSolaris development community requesting a contact to ensure that proposed changes to OpenSolaris would not break OpenAFS. [13:22:50] for something that significant, I would expect to see a subject line saying something to that effect to *-announce and -devel. [13:23:04] (ie, I'm one of those folks that don't read every single message) [13:23:17] s/expect/hope/ [13:23:23] if you look at the thread I think there was one [13:23:38] I just searched the past 3 months of -devel..didn't see it. [13:23:47] It's older than that. [13:24:45] I specifically remember an invite to join message after the discussion as to whether there should be another port list or not. [13:25:28] http://lmgtfy.com/?q=openafs+solaris+porting+list [13:26:03] * steven.jenkins thanks russ. [13:26:29] not sure why/how I missed that, but I did. [13:43:43] --- deason has left [13:44:13] --- deason has become available [13:48:53] > for something that significant not significant enough for announce, by far [13:49:17] if you don't already know you care about maintaining the solaris port and i have to announce it to you, you've proven you still won't care [13:52:52] (because the noisier we make -announce, the more people will decide they don't have time and punt.) [13:53:25] anyway, if a message titled "port-solaris list created" to openafs-devel isn't enough, uh, i dunno. [13:54:03] I know what you mean. but I know a developer who might be interested in the port-solaris list but who would not be interested in the higher traffic openafs-devel. [13:54:27] ie, he's not on openafs-devel already because of the volume, so he wouldnt have seen the announcement. [13:54:38] now, of course, mea culpa for not seeing the announcement in the first place... [13:54:47] which is precisely the situation simon hates, namely, wants to live in a walled garden and repeat mistakes the general community has already solved. [13:55:45] I'm looking at it quite differently: he's very very interested in solaris internals, but not so interested in afs. this list might be a way to get him more involved in openafs development. [13:56:09] I agree w/simon's concern, btw. I'm not a big fan of splitting up the dev lists. [13:58:09] arguably it should have been in the november newsletter. none of us apparently thought about it [13:58:56] not a huge problem -- now that I know about, I'll pass the info along. [15:05:20] --- jaltman has left: Disconnected [15:05:30] --- jaltman has become available [15:33:09] --- jaltman has left: Disconnected [15:40:37] --- deason has left [16:22:37] --- deason has become available [17:17:14] Actually, I'm not sure that I'm on any openafs lists as kaduk@mit.edu right now (maybe not even port-freebsd, though I seem to recall that that doesn't get traffic). Which ones do I need to be on to keep myself from wanting to kick myself later? -devel, -info, and -testers? [17:32:28] if you care about the ports lists, the non-darwin ones may get traffic someday [17:33:54] it doesn't look like -testers has had traffic in almost six years [18:06:43] i assume port-freebsd will get traffic again [18:10:05] Yeah, if we are going to actually have three people other than you working on it, individually-addressed emails are probably poor. [18:10:46] While you're here, which makefile(s) do I want to tweak to get debugging symbols? My first attempts have been clobbered ... [18:11:13] Assume I've already run configure. [18:16:02] agreed. [18:16:16] makefiles: don't. just use the configure options. [18:16:27] if you must, tho, use src/config/Makefile.config, after running configure [18:49:02] Why ... does rxfs_storeMemOps not have a .storeproc entry? [18:52:59] oh. it didn't need it but then chas williams rewrote it without understanding and i fixed one case but not all [18:54:05] Do you want that I should submit the patch? [19:03:07] Oh, hey, writing files doesn't hang anymore ... [19:04:03] ... sometimes? [19:06:33] sure. it should be a one-liner [19:06:54] Ah, I bet that my first test was already in local cache, and didn't have to wait for the network :( [19:07:09] Well, there is the __c99 and non-__c99 case ... [19:07:10] yeah. the network issue is still there [19:07:42] You have been poking at it? [19:08:31] Also, this aklog abort is getting pretty annoying. Maybe I should go track it down. [19:08:32] i've said it at least 3 times now... [19:10:04] The last I remember is: -> openafs@conference.openafs.org / / shadow@gmail.com/owl7CDEA696 i have the freebsd vm in front of me, the wakeup issue... i want to look at a macos issue then i will go back to it. i have an idea as to the issue [19:48:32] --- mho has left [19:48:33] --- abo has left [19:49:16] --- abo has become available [21:18:55] --- jaltman has become available [21:33:12] --- tharidufernando has become available [21:41:02] --- tharidufernando has left [22:04:41] --- reuteras has become available [22:21:43] --- deason has left [22:27:20] --- jaltman has left: Disconnected [22:28:52] --- jaltman has become available [22:43:01] --- Jeffrey Altman has left: Disconnected [22:47:11] --- jaltman has left: Disconnected [22:48:13] Is there a story behind /afs/grand.central.org/software/openafs/git/openafs.git ? [22:48:35] i'd bet there is. i probably even know it [22:48:43] i'm not being coy. i don't remember [22:49:31] fwiw, aklog. are you on amd64? [22:52:09] Yup. [22:53:12] It is seriously messing up the stack and/or heap somehow, it seems. I will try running it in gdb with malloc debuggin options on at some point, but otherwise it looks like I'll have to bisect it. [22:54:07] (If that openafs.git isn't going to get updated, it might be worth thinking about removing it instead.) [22:57:57] i'd bet if you enable lwp debug (compile lwp with -g) the crashing stops [22:58:54] So, I did break down and re-configure with --enable-debug. Not sure whether that gets passed through to everything or not, though. [23:00:19] --- kaj has left [23:01:05] i suspect you want --enable-lwp-debug or --enable-debug-lwp whichever it is [23:02:02] Ah. [23:28:38] --- summatusmentis has left [23:51:57] --- kaj has become available