[00:01:00] --- reuteras has become available [04:00:15] --- mmeffie has left [04:39:22] --- Simon Wilkinson has become available [05:21:37] --- SecureEndpoints has left: Replaced by new connection [05:39:30] --- mmeffie has become available [05:50:32] --- SecureEndpoints has become available [06:00:45] --- Simon Wilkinson has left [06:07:19] --- haba has become available [06:15:11] --- Simon Wilkinson has become available [06:17:39] --- Simon Wilkinson has left [06:19:29] --- Simon Wilkinson has become available [07:06:37] --- Simon Wilkinson has left [07:06:37] --- Simon Wilkinson has become available [07:07:04] --- Simon Wilkinson has left [07:07:58] --- Simon Wilkinson has become available [07:14:52] --- Simon Wilkinson has left [07:14:52] --- Simon Wilkinson has become available [07:41:28] --- Russ has become available [08:36:11] --- matt has become available [09:31:04] CentOS 5.3 release is set for 29/03. I'll get OpenAFS RPMs built as soon after that as I can... [09:31:21] thanks [09:32:20] --- mmeffie has left [09:52:19] Just saw this link in another jabber windows: http://www.openoffice.org/issues/show_bug.cgi?id=100542 (AFS is mentioned in the bug report) [09:53:18] Short summary: File is unlinked when quota full. [09:53:34] Yeh. Not AFS specific. Their error handling just sucks. [09:54:09] Of course not AFS speciffic, but I allready see the blaming. [09:54:21] Hi Simon btw and welcome back [09:54:27] Hi haba. Thanks! [09:55:09] everything is afs' fault [09:55:21] No. Everything is LDAP's fault. [09:55:24] --- summatusmentis has left [09:56:11] even when you can prove otherwise. 's mgmt still blames AFS even when accepts responsibility for the bug. If it wasn't for AFS that bug would never have been hit. [09:56:51] makes me want to retire to Belize with my puppy [09:57:51] * haba wikipedia:ed Belize [09:57:54] I think we need to parametrise [10:03:07] Jeff: (after looking at the pictures of Belize) mee toooooo. [10:04:10] Simon: That Centos RPMs, would that be 1.4.8 as is or with "vendor patches" of bugfixes so far? [10:04:51] haba: Good question. [10:05:08] We don't have to take any patches for kernel compatibility. [10:05:22] We should perhaps take the rx idle dead patch. [10:05:50] I'm nervous about using the RPM packaging as a way of doing sub-releases though. [10:06:13] (where those subreleases are addressing more than just what's necessary to deal with kernel builds) [10:06:19] Anyone else got thoughts on this? [10:06:44] I have some patches in the openafs I install on my centos boxes. Some are PDC specific choices but through the years I have often had one or two bugfix patches in the rpms I did for PDC. [10:06:51] Debian packages diverge quite frequently [10:07:19] I know, but they're Debian packages, not OpenAFS packages. I think that distinction is important. [10:08:13] --- stevenjenkins has left [10:09:09] that is true [10:09:44] I don't want to end up with a situation where the 1.4.8 (say) in our RPMs is hugely divergent from the 1.4.8 that's in the shipping tarballs. [10:09:58] 1.4.9 will be out shortly [10:10:14] I'm happy to take patches that are necessary to deal with changes in the OS. I'm not (yet) convinced that I should take patches that fix bugs in 1.4.8. [10:10:20] we have in the past issued "a" versions. [10:10:26] Indeed. RPMs for pre1 are building as I speak. [10:10:43] The only bugfix patch I have currently is STABLE14-rx-idledead-only-ignore-keepalives-20081222 [10:10:56] Yup. We've got that locally now, too. [10:11:11] --- summatusmentis has become available [10:11:24] --- summatusmentis has left [10:11:26] if you believe that it is critical for us to have 1.4.x release packages for Centos5 then we can tag 1.4.8a with that one patch and you can build off that tag [10:11:35] The I patch src/config/param.ia64_linux26.h to use USE_UCONTEXT [10:12:04] haba: If you have other patches that you think are relevant, why not open them as bugs in RT. That way we've got some kind of record of them. [10:12:09] --- stevenjenkins has become available [10:12:19] --- summatusmentis has become available [10:13:04] It has been quite a while I built on ia64 and the patch is from 2007-04-25 [10:13:14] but yes... [10:13:49] SecureEndpoints: I think, given that 1.4.9 is imminent, we can avoid tagging on this occasion. [10:15:47] * haba should try to build on ia64 this week [10:19:35] --- dev-zero@jabber.org has become available [10:19:40] --- dev-zero@jabber.org has left: offline [10:43:17] * haba done for today [10:43:27] *wave* [10:43:41] --- haba has left [10:51:34] --- reuteras has left [11:21:21] > src/config/param.ia64_linux26.h to use USE_UCONTEXT this is probably correct [11:26:18] --- Simon Wilkinson has left [13:36:44] --- mdionne has become available [13:41:00] how imminent is the 1.4.9 release looking? it would be nice to include patches for linux kernel 2.6.30. rc1 should be in about 2 weeks [13:45:07] at least a week off. [13:45:25] and depending how invasive those patches are, shipping them this late woudl sort of suck [13:45:43] I only know of 1 (minor) issue so far, based on linux-next. but that's not always an accurate predictor [13:48:09] i'd be surprised at this stage to see anything invasize show up that wasn't in linux-next [14:31:58] --- mmeffie has become available [14:44:41] --- summatusmentis has left [14:56:05] --- summatusmentis has become available [15:57:25] IMHO, if it's a bug in RHEL5 (CentOS5), then, a bug report should be submitted to RedHat [15:58:10] that said, if RedHat is going to take "too long" to fix said bug, a workaround in afs would be nice... but... it should be in the main source, not just in the rpm [15:58:24] er. what? [15:58:33] w. regards to the earlier conversation [15:59:03] hrm... Simon's not hear [15:59:06] er, here to hear [15:59:22] so, n/m [15:59:38] it wasn't obvious at that point what bug you meant either [15:59:46] and looking at the log it still isn;t [15:59:53] well, I am responding to smething talked about 5 hours ago [16:00:04] I have no idea what the actual bug is (and at the present time don' care since I use RHEL4) [16:00:12] but, it's the principle of the thing :) [16:00:55] i read the log, it covered 5 hours ago. i still have no idea what you're talking about [16:01:52] Simon said... I don't want to end up with a situation where the 1.4.8 (say) in our RPMs is hugely divergent from the 1.4.8 that's in the shipping tarballs. [16:01:59] yup [16:02:06] which still refers to no specific bug [16:04:22] looks like itgrew out of kernel compatibility, rx idle dead patch, and a couple of other things haba mentioned [16:05:19] btw, what does OpenAFS expect or do as far as a development environment for its GSOC participants? [16:07:18] it depends on the project. we support many different environments. [16:07:56] not as in some IDE, but as in a machine to compile stuff and test stuff on [16:08:08] you (as mentor) pretty much get to determine which platform you wish to support [16:08:31] so as a mentor, it's my responsibility to provide a machine for the student to play on? [16:08:41] we expect students to have a machine [16:09:35] if they don't have a machine, how are they going to connect to yours? [16:09:49] a machine or a machine to use as a development machine? [16:10:02] who is expected to provide the development machine(s)? [16:10:08] they are [16:10:17] k... as best I can tell, that isn't stated anywhere [16:10:22] what dev tools require a dedicated machine? [16:10:53] depends on the load on said machine and whether or not what you're working on requires root [16:10:57] we also don't say we're giving access either. [16:11:10] it's a question propsective students have asked me [16:11:38] so far I've had to say "I don't know" since I couldn't find any documentation [16:11:57] we do not make any commitment either way [16:12:02] OpenAFS has no hardware to give [16:12:16] we can advise on how to setup a linux VM [16:12:25] if the expectation is that students provide their own hardware, that should be stated [16:12:30] we can advise on how to compile on their MacOS machine [16:12:48] the expectation is they'll talk to us [16:12:52] no one else in GSoC does to my knowledge [16:13:09] --- abo has left [16:13:09] i guess we should state that they should talk to us? seems kinda implicit. [16:13:10] --- stevenjenkins has left [16:13:12] some orgs have VM images they provide [16:13:16] you should state that [16:13:28] it depends entirely on the project. [16:13:29] --- abo has become available [16:13:36] what are the requirements for your project [16:13:40] ? [16:13:49] it depends on the plan that is developed [16:13:59] since the first part is to develop a plan [16:13:59] i'd actually argue a student who doesn't ask questions is probably not the best one to adopt. [16:14:31] sometimes it's less about not asking questions and more about asking the right people, or being confused because one is still a sophomore, or being shy, or whatever [16:14:33] if you are developing a web site that is very different from developing a user mode shell extension on Windows vs a finder extension on MacOS vs kernel development vs a distributed database extension [16:14:46] like I said, I've been asked the questions [16:15:00] then send them to the IRC channel or this forums [16:15:05] let them ask [16:15:13] have done so... maybe they will come in, maybe not [16:15:23] one is stopping by to talk to me tomorrow or Friday [16:15:24] that is all you can do [16:15:38] true... though, "pass the buck" also looks bad [16:15:44] a student that won't interact with the community is not going to be a good fit [16:15:54] that is not pass the buck [16:15:57] you are not the admin [16:15:59] --- stevenjenkins has become available [16:16:13] when your users report bugs in software, do you expect to fix them youself, or do you pass the buck? ;) [16:16:15] and we are perfectly happy to help good students [16:16:31] its a case by case basis [16:17:04] I'm not going to fedex a laptop to a student that I don't trust and I'm not going to promise that I will or won't [16:17:23] I'm just saying, if asked some of these questions, i'd like to be able to answer rather than say " go talk to the community" [16:17:29] which is why I'm asking now :) [16:17:39] the answer is "go talk to the admin(s)" [16:18:05] or you are welcome to offer your own resources if you have them. [16:18:32] honestly, I don't really [16:18:41] basically it boils down to "if there is a good student, we will figure something out, but that something might be their CS dept giving them resources" [16:19:06] seriously its a case by case basis [16:19:10] ok [16:19:18] good to know [16:19:27] if the student is local to nyc I have no issue will giving them a desk to work at and a machine [16:19:32] --- abo has left [16:20:05] sure [16:20:30] --- abo has become available [16:27:26] there is a small possibility I could help locate some basic resources (e.g., relatively small, but functional, VMs running CentOS) [16:34:43] I suppose it's all a moot point until students are accepted [16:39:27] It's challenging to use an openafs development environment once you have one. [16:40:27] how so? [16:41:06] Just the challenge of understanding all the components. [16:41:09] Of afs itself. [16:41:13] ah, yes [16:50:39] one of the most common strategies is one of the toughest: 'diving right in, installing, and trying to get your head around everything'. [16:50:56] Its the way to win, for the winners. [16:51:14] it's a rather harsh Darwinian method, though. [16:52:20] that method isn't really about the skills of the mentors but much more about skill + persistence of the student (and I use 'student' in a broad sense here to mean anyone new to AFS) [16:53:16] I'd like to be a great mentor, with students who would really dive in and get traction. [16:55:51] ditto. [16:57:02] several yrs ago, I told one of bosses that I was tired of teaching graduate students and would like to teach Freshman/Sophomores for a while, as I had burned out somewhat in trying to handle the lack of desire to learn on the part of most of my grad students [16:57:09] err, 'one of my bosses'. [16:57:39] so I taught an 'Intro to Unix + shell scripting' class for a while..turned out to be a ton of fun. [16:57:41] "lack of desire to learn on the part of most of my grad students"--are you serious? [16:57:47] completely. [16:57:53] That is deeply saddening. [16:57:59] yes, very. [16:58:31] I had it from undergrads--I never let myself imagine it went higher. [16:59:38] When you feel that about your work, it's time to do something else. [17:00:35] typical attitudes: large sense of entitlement, view of grad school as a strictly professional degree, so had a desire to minimize difficulty and maximize 'good grades'; constant pressure to lower rigour so as to widen the potential pool of students, etc. [17:01:22] I like teaching a lot, but I definitely needed a break from the grad program. [17:07:12] Do you think you'll return to academics at some point? [17:07:46] hard to say. That was one of the toughest questions Jon (our CTO) asked me during my interview ~2 yrs ago. [17:08:28] I'm embarrassed to say that I blew off finishing my PhD right before my last job, which kind of burned some bridges.. [17:09:02] I had about 4-6 months of work left before my dissertation defense..and skipped the country. :/ [17:09:32] All you missed was the defense? [17:09:49] oof... can I ask what school? [17:09:55] passed qualifiers, passed prelims, had the research started..then split. [17:10:12] was well into the research,actually.. [17:10:16] Iowa State [17:10:45] I left there in the mid 90's, but petitioned them circa 2003 to let me finish...then blew them off again. [17:10:50] ouch [17:11:03] yeah. like I said, I kind of burned some bridges there. [17:11:17] no hard feelings anywhere, but it would be hard for them to let me finish at this point. [17:11:49] were you a cs student? [17:11:52] otoh, if I showed up w/my dissertation written, I might beg. However, my major prof left there ~2 yrs ago, so it would be tough. [17:11:54] yes. [17:13:02] --- abo has left [17:13:49] --- abo has become available [17:15:17] unless you want a PhD research position or want a tenure track position or want the accomplishment for yourself, there is little in this world that requires a CS PhD. when it became clear to me that I did not want any of those I walked away from mine. [17:17:21] * stevenjenkins nods. I was in a tenure track position when I left academia. Certain aspects are good; certain ones aren't so good. [17:18:47] conceptually, I like the idea of a research position [17:20:23] that's interesting. the first time I left grad school (mid 90s) was because a research position was too confining, too narrow for me. [17:21:11] In many fields you need a big organization in order to provide you the resources necessary to conduct research. With CS, you can do it from a beach or your backyard. [17:21:26] all it requires is your imagination and persistence [17:21:39] I wouldn't mind doing research, say, 2 years out of every 5, but that's probably all I could handle w/o my wife wanting to strangle me because I get too distracted/embedded into problems and forget to do basics like talk to people, go outside once in a while, etc. [17:22:12] it's concievable that I'll find a research position too confining, not there yet [17:22:18] the monastic/cloistered lifestyle was the biggy for me. [17:22:23] that is why we now have a puppy. my wife wants me to be forced to leave the office to walk him. http://www.sara-jeff.info/our_puppy/ [17:22:28] heh. [17:22:49] we recently got a puppy as well, although we have 3 kids, so my life is a bit busy w/them already. [17:26:34] puppy good [17:27:10] are you bringing leo to the bpw? [17:30:48] don't know about leo and bpw. I doubt I can keep him at the guest house [17:31:09] I am taking him to DogFishHead brewery in early May [17:31:17] that and all the way across the country :) [17:32:05] flying is easy. just costs $150 each way [17:32:55] oh, beagle, right [17:33:12] my family has always had pyrenees or newfoundlands, can't easily fly with them [17:53:02] --- matt has left [18:05:55] --- mdionne has left [18:07:31] --- mdionne has become available [18:38:29] --- mdionne has left [18:44:09] is there a way to get server proximity and bandwidth throughput directly from rx peer stats? [18:44:53] or if not, is there something that'll tell me time lapsed for the packets sent/resent bytes sent/resent stats? [19:13:10] proximity: no. you might be able to augment peer to keep rtt if it doesn't and derive info from that. but it's not true proximity, you actually want to keep like... first derivative of rtt so it's smoothed when there are horrible effects from brief saturation of a network, i'd bet [19:14:37] rtt? [19:16:13] round trip time [19:16:35] oh [19:17:07] d'you know if peer does that currently? [19:17:20] no, but i bet rx.h will tell you [19:17:41] was just checking out src/rx [19:17:50] int rtt; /* Round trip time, measured in milliseconds/8 * / int rtt_dev; /* rtt smoothed error, in milliseconds/4 */ [19:18:30] awesome, thanks [19:18:52] I may have more questions later :) [19:19:28] ok [19:21:29] delayed: i'd enjoy having a research position as long as it wasn't narrowlyfocused. [19:42:21] derrick, why /8 and /4? [19:45:10] probably historical [19:45:32] no need to keep such accuracy if your networks are significantly slower [19:46:39] it was like that when i found it [19:46:52] ah. hrm [21:30:29] --- Russ has left: Disconnected [22:29:55] --- dev-zero@jabber.org has become available [22:32:22] --- dev-zero@jabber.org has left [22:49:25] --- reuteras has become available [23:51:03] --- SecureEndpoints has left: Replaced by new connection