[00:52:17] --- lama has become available [03:58:02] --- abo has left [04:13:11] --- abo has become available [04:40:14] --- lama has left [05:27:12] --- lama has become available [07:14:46] --- deason has become available [07:27:19] --- reuteras has left [09:09:05] --- Simon Wilkinson has become available [09:30:10] Is more testing of 1.6.0pre2 to try and track down vos issues useful, or should I wait until pre3? [09:31:26] --- rra has become available [09:32:49] Given the nature of the ulvldbentry / nvldbentry revert that just went through, I'd be inclined to wait till pre3 [09:33:12] funny you should ask. tyou're on release-team, right? [09:33:20] I should be. [09:33:29] As should I. [09:33:32] But no mail. [09:33:59] Mind you, I stopped getting mail from RT months ago, so maybe grand just hates me. [09:34:34] i (had) just clicked send, so.. [09:36:28] What's the issue with things without O_LARGEFILE? I saw Andrew's comment, but didn't understand it. [09:36:46] If we don't have O_LARGEFILE, we use preadv and pwritev, and we should work anyway, surely? [09:37:45] well, the configure test should technically only disable piov if O_LARGEFILE is defined, right, FDH_mumble only use them if O_LARGFILE... [09:38:38] pretend i am not typing badly. sun got in my eyes [09:38:42] Oh yeah, the configure test could be smarter. But I figured it was more important that we actually compile at that point :) [09:42:37] I also don't know how likely it is to have preadv and no O_LARGEFILE, so eh... [09:42:55] yeah, that was my thought [09:43:13] (as i commented) [09:44:55] i am sort of tempted to push a patchset with a configure test we subsequently junk which just tells us on which platforms we have what support for 64 bit filesets properly with file offset bits et al [09:45:17] I suspect that we can rely on autoconf to get that right for us. [09:45:43] What we need to check is that we don't care that off_t suddenly becomes 64bits wide [09:46:01] And then we can throw away all of the O_LARGEFILE based checks [09:46:13] true [09:46:54] some things break on platforms when you enable 64-bit wide off_ts, though; glibc preadv was apparently broken for a short time with _FOB64, and hpux's pio is all broken with it [09:47:41] so, there is no way to turn on 64-bit off_ts and "get it right" everywhere [09:47:51] er, automatically via autoconf, I mean [09:48:11] How widely tested has the positional I/O stuff actually been? [09:50:57] it hasn't been in production anywhere recently; awhile ago it might have been, but I'd guess derrick might remember better than me [09:52:19] I guess my question is - "are we confident in our preadv/pwritev code, and in the solidity of kernel support for it that we should ship with it enabled in 1.6.0?" [09:52:49] if we are turning it off, i'd sooner just never define HAVE_IOV than anything else [09:53:14] HAVE_PIOV ? [09:53:20] that. [09:53:24] yes, pulling out the code itself I think would be a bit painful at this point [09:54:11] I'd think platform issues would be spotted rather quickly, though [09:54:15] Sure, disabling it seems like a better bet, if we think we're going to turn it back on later. [10:08:04] is there anything like 'git cherry', but works off of "cherry picked from commit" lines in the commit message instead of via patch ids? [10:08:40] Yeah. I wrote something. [10:09:04] Given two branches, will list everything that's on one branch but hasn't been pulled up to the other. [10:18:25] --- Simon Wilkinson has left [10:41:18] --- Simon Wilkinson has become available [11:44:35] --- Simon Wilkinson has left [12:03:23] --- sxw has become available [12:06:50] --- mdionne has become available [12:13:26] --- sxw has left [13:00:18] --- sxw has become available [13:08:06] --- sxw has left [13:14:22] based upon the current level of mentor response, I will not be filing an application for GSoC on behalf of OpenAFS this year. [13:18:31] How many responses do you have? [13:30:53] none [13:32:00] Andrew and yourself have indicated that you might be available but no firm commitments [13:33:52] I don't know how I could have given a more firm commitment than I did [13:35:04] I can be available, if needed. I had gotten the impression that there were other willing mentors for the projects I knew about. [13:35:13] I literally said "yes, I commit" [13:35:19] And Andrew sounded pretty firm to me. :-) [13:35:44] then put together a project list and find others. [13:35:57] I won't submit an application with less than four mentors [13:36:02] but still, just jhutz and me is still only 2 [13:36:04] yeah, that [13:37:02] just to confirm, the two of you are prepared to spend the 30 hours a week during the two weeks of the application period responding to student inquries in the IRC channel? [13:37:37] Which two weeks are those? [13:37:48] you can check the schedule [13:39:09] http://www.google-melange.com/document/show/gsoc_program/google/gsoc2011/timeline it runs March 18 to April 8 this yer [13:40:32] yeah; I just found that page. [13:41:20] nobody mentioned anything about an additional time commitment during those two weeks [13:42:16] oh wait, yes you did [13:42:17] :) [13:42:24] uh huh. [13:42:24] just not a specific amount of time like that [13:43:06] if there are many mentors in multiple time zones the load becomes smaller [13:43:17] in past years we have had 8 to 10 mentors [13:43:18] wait, no, I read it wrong, still [13:43:26] "As those that have mentored in the past can attest, mentoring is a significant time commitment starting during the student application period the week of 28 March 2011 and ending on 26 August 2010 if you are lucky enough to end up with a student. A mentor should expect to spend ten to fifteen hours a week working with her/his student and the other accepted students." [13:44:13] That's actually three weeks, because it includes the one week discussion period before applications open, but that doesn't bother me. I'm certainly willing to hang out in IRC and talk to people about their potential AFS projects. I think that means I need a newer barnowl. :-) [13:44:39] i know i need a newer barnowl, but i have no time to dink with johnstown just now [13:47:50] honestly, the irc requirement bothers me more than the time requirement; but the point is moot if we're not going to submit anything without more mentors, anyway [13:50:12] i hate irc. [13:50:52] I've always hated irc [13:59:38] but yes, I _can_, to answer the question [14:10:36] hmm, if the afs3-stds archive threading was fixed before, it's broken again [14:11:26] that is, the [Thread] et al pages all point to the same message, and jan/feb 2011 don't show up (though, again, the .txt{,.gz} archives seem fine) [15:15:03] --- deason has left [15:28:31] --- Jeffrey Altman has left: Disconnected [15:32:35] --- Simon Wilkinson has become available [15:59:15] --- deason has become available [17:46:01] --- rra has left: Disconnected [18:05:42] --- Russ has become available [19:08:36] --- mdionne has left [22:18:03] --- deason has left [22:52:44] --- reuteras has become available [22:57:31] --- lama has left [23:46:25] --- Russ has left: Disconnected