[11:31] danilos, oops :-P [11:31] sorry and thanks [11:37] gary_poster, does the change look good? I wasn't sure if I was just not seeing something [11:38] danilos, actually...that's not quite what I want, but what I want may not work for make harness. [11:38] lemme actually merge your branch [11:38] gary_poster, sure... I just re-introduced the two lines that were there previously at roughly the same spot [11:39] right danilos; my intent was to move them down into success/failure; that means that you get start and stop times like you do with the OOPS report, which I found nice. [11:39] that may not be make-harness-friendly though [11:43] danilos, no, what you've done effectively doubles what I've done, as I feared. We may need to rip out one of the changes I made (an improvement, IMO, so that's too bad) to make it work as you want [11:43] danilos, take a look at lines 685-687 in the file in your branch [11:44] yay! sweet, sweet Internet [11:44] benji, hiya! So, another quake yesterday evening? I got your mail. What was the Richter scale number? [11:45] gary_poster: heh, no, different natural disaster this time, major thunderstorm [11:45] oh, ok benji [11:45] gary_poster, oh, indeed [11:45] trees blocking the road, power lines down [11:45] oof [11:45] gary_poster, I also had trouble with scripts (where I noticed it first) [11:45] on my ride this morning it looked like we were hit with a huricane, but no, that's tomorrow [11:45] danilos, that really seems odd, doesn't it? [11:46] yeah, benji :-( I hope that's not as bad as it might be [11:46] for the whole east coast [11:46] if the rivers turn into blood I won't be working for a few days [11:47] gary_poster, well, I think it's quite obvious, action is None in there, perhaps we can just log the SQL statement if action is None? [11:48] gary_poster, of course, ideally, the tracer would be installed for the script layer as well [11:49] danilos, right, but how could action be None? and presumably the tracer must be installed for the script layer if it worked before [11:49] danilos, notice that the previous stderr.write that you reinstated in your branch is in the tracer [11:49] but before the SQL is called [11:49] gary_poster, yeah [11:50] and I simply have it in the same tracer, after the SQL is called [11:50] and the code before the SQL is called stashes the action on the connection in all cases [11:50] gary_poster, if request_starttime is None: [11:50] return [11:50] gary_poster, that could be the culprit [11:50] danilos, ah, duh, of course [11:52] danilos, http://pastebin.ubuntu.com/675182/? [11:52] danilos, except we want the return too :-P [11:52] gary_poster, yeah :) [11:53] http://pastebin.ubuntu.com/675183/ [11:53] gary_poster, btw, would it perhaps be possible to record the start time as well or do we really need a request? [11:53] gary_poster, (really need a request to be able to use the same code path: it'd be nice to get timings for running scripts as well) [11:54] gary_poster, otherwise, r=me :) [11:57] gary_poster, fwiw, if you don't want to be bothered creating a branch with this, I've pushed the suggested change up, you can ack it so we can land it, but if you want to fix it in a better way, you can still do it :) [12:01] :-) danilos, "record the start time": we could get a duration easily enough, I guess, by scribbling the current time and the statement_to_log on the connection when we don't have an action; or we could do mildly tricky things with the transaction module so that a commit or an abort resets a connection scribble of start time, and "0" is always guesstimated as the first SQL request of the transaction. [12:02] I don't want to do the second idea; the trickiness is non-trivial (relatively easy and fun to code, but things could go wrong in potentially serious ways so I'd want tests and stuff). [12:02] The first idea would be ok. looking around for a sec... [12:02] danilos, I have another related branch to fix another sqlprofiler issue that I was unable to land because of some ec2 image problems I had last night. I can just combine and land them. [12:09] gary_poster, ok, sure, I'll scrap my branch then [12:10] gary_poster, I hope it's not the versioninfo thing, that was supposed to be fixed by now :/ [12:11] danilos, you mean my ec2 problems? I think it was just because I was using a version of ec2 test from before StevenK blessed himself as an image provider, so I got an earlier revision [12:11] If not, I'll be surprised [12:11] gary_poster, oh, ok [12:16] gary_poster: oh, Bella is going on a play date today, she's being picked up around 8:30 so I may disappear from the call [12:17] ok benji. as long as that doesn't involve you disappearing down into the crevice of an earthquake's fault line, or up into a hurricaine's sky, I'm OK with it. [12:18] heh [12:18] I expect locusts at any moment. [12:19] :-) [12:31] bac benji danilos gmb, call in 1 [12:33] danilos, http://pastebin.ubuntu.com/675206/ [12:40] gary_poster, looks nice, albeit 0-something for every query is a bit dull, but good enough for me :) [12:46] bac: screen sharing is easy if both users have ssh access to the same host, is that the case? [12:46] gary_poster, oh, btw, I hope you'll be able to re-land your change using getMergeProposals() today, but the same caveat holds for you: you might need to pass in visible_by_user to get the private branches as well [12:47] gary_poster, (my branch is in ec2 with the private branches fix, basically just adding a test and that parameter to the call) [12:48] danilos, 0-something: you want me to put in an XXX? I agree that it would be nicer, but not enough to do what I can see would fix it. [12:49] JS question: If I load some HTML containing JS via an async request, that JS won't automatically be executed, will it? (e.g. pulling in another page via Y.io won't load that page's JSON object cache, will it?) [12:49] gary_poster, I don't even care about the XXX [12:49] gmb, nope, it'd have to get "rendered" first, what you pull up is just text [12:49] gmb, if it is json, some of our own methods will parse it though [12:50] gmb, lp_client.named_get or something like that [12:50] danilos, ok cool. :-) re-land my change: ok I'll take a look, thanks. I suspect that for my case I won't need to because I'm already passing in branches that have been filtered in that way [12:50] benji: i guess it can be the case, if we're willing to grant access to each other [12:50] gary_poster, right, if the call was already getMergeProposals() before, you should be good [12:51] benji: so one just creates a screen and the other connects to it? [12:51] bac: basically. you have to turn multiuser on and grant the other user access, let me see what I can find on the net... [12:52] benji: that's ok, i'm still looking [12:52] bac: here's a good description: http://stackoverflow.com/questions/471354/sharing-a-gnu-screen-session/1896576#1896576 [12:53] danilos, cool [13:00] danilos: Thanks. [13:32] benji, bac, my concern is that I believe we want ssh tunnels involved too, and I don't think that's what is going on in the stackoverflow answer (and I think that's the interesting part of what the LOSAs know how to do). I suspect it is easy...ah. [13:32] https://wiki.canonical.com/InformationInfrastructure/OSA/ScreenSharingWithDevs FWIW [13:33] gary_poster: other than using ssh to connect the terminal session, what tunneling do you mean? [13:34] that's all I mean, benji--what that page describes is what I was hoping for--actually, more paranoid, as you'd expect [13:35] (the "more paranoid" part is that I'd expect we'd just let the other person act as us. Maybe we ought to include that paranoia too, though.) [13:36] gary_poster: benji and i are working on tunneling now [13:36] yay, cool [13:40] gary_poster: in that wiki page i don't understand the step: "Have the dev login as the user directly" [13:44] bac, like, as the pgsql user or the mongo user or what have you [13:44] IOW something specific to how deployment machines are usually set up [13:45] Some of that happens on dev machines, but we pretty much do everything as us [13:50] * gary_poster going to do some CHR, since I've not had a good week for it. [13:57] IT WORKS, LET's MERGE IT. [13:58] Anybody ever heard of Launchpad Results? [13:58] I mean... oh, that's nice. [13:59] all, I am off, and I forgot to mention that I am off on Monday: enjoy the weekend all :) [14:00] gary_poster: Never heard of it until now. [14:01] gary_poster: looking at the code, it seems to do a lot of reinventing the wheel... [14:02] Ah, maybe that's because they want to represent LP stuff in a non-LP webapp/microsite/somethign. [14:19] bye danils [14:19] danilo_ that is [14:36] hi benji, i got disconnected [14:36] I figured. :) [14:36] benji: you should be able to get in now, right? [14:36] it worked! I had to specify the user name, like so: ssh guest@spoetzl.net -p 2200 [14:47] cool [14:49] thanks for the help benji. i'll write this up [14:49] my pleasure [14:50] gary_poster: did you get my question earilier about the wiki instructions? [14:50] bac, you mean "gary_poster: in that wiki page i don't understand the step: "Have the dev login as the user directly""? [14:50] gary_poster: yeah. sorry, my connection dropped and i lost irc history [14:51] bac, np. This was my reply: [14:51] bac, like, as the pgsql user or the mongo user or what have you [14:51] IOW something specific to how deployment machines are usually set up [14:51] Some of that happens on dev machines, but we pretty much do everything as us [14:52] bac, make sense? I'm about to run out to an early lunch [14:52] gary_poster: no [14:52] gary_poster: we can chat after lunch [14:52] bac, admins run processes as different users [14:52] so there will be an apache user [14:52] a lp user [14:52] a postgres user [14:52] and so on [14:52] gary_poster: right, i understand that [14:53] so the dev should log in as the user that they need access to [14:53] that's what I understand [14:53] gary_poster: so do the losas give you an account and passwd for, say, the apache user? [14:54] gary_poster: again, no rush. enjoy your lunch. [14:55] no, bac, not an account. they put the dev's key in the user's authorized keys. See step 2 on that page [14:55] "[...] to (as example) marang:/etc/ssh/user-authorized-keys/couchdb" [14:56] so the dev key is now a temporarily authorized key for the couchdb user in that case [14:56] * gary_poster goes now bac. :-) [14:56] talk to you in a bit [15:02] hey benji, try logging in as guest@spoetzl.net again. this time it should ask for your SSH credentials [15:04] bac: it's prompting me for a password (and not accepting the one that worked before) [15:04] benji: try your ssh credentials password [15:04] or is it wanting a unix password? [15:04] it's wanting guest's password [15:04] (unix) [15:04] benji: i tried using the steps here https://wiki.canonical.com/InformationInfrastructure/OSA/ScreenSharingWithDevs [15:05] and entered your ssh key from LP for guest [15:05] i wonder if there is sshd config needed to allow user-authorized-keys [15:05] * bac re-googles [15:19] bac: when you get a second, will you look at https://code.launchpad.net/~benji/launchpad/bug-820055/+merge/73064 for me? [15:19] benji: sure [15:19] I'm going to lunch now. [15:19] thanks [16:10] bac, I'm back, but it sounds like you're now past what we were talking about [17:04] gary_poster: yes, thanks. [17:05] cool [17:31] * gary_poster tried to pay penance for his poor-CHR ways this week, but is now finished. [17:38] gary_poster: work in progress https://dev.launchpad.net/yellow/UsingScreen [17:43] bac, nice. I'm still trying to understand all of it. [17:43] gary_poster: it's still sort of hacky. don't like having to pass around passwords [17:43] and the command limit is nice. but it doesn't work atm [17:45] bac, passing around passwords: I take it that getting the key from Bob and putting it in authorized keys didn't work? That's what I would have expected to be the answer [17:45] I've seen the command limit work, of course, but I'lve never set it up myself [17:45] gary_poster: yeah, that'll probably work if i remove the 'command=' part [17:45] huh [17:45] right now it reports "no screen to connect to" [17:46] fwiw, the combination (both command and authorized keys) is how, for instance, Zope.org svn was and probably still is set up [17:46] so "it ought to work" but I know that's not particularly helpful [17:51] not yet, but I may when my current ec2 run is done [18:08] benji, I suspect that's a follow-on to my ec2 woes in #launchpad-dev? [18:08] heh, yeah [18:09] * benji resets the "It's been 14 days since an IRC channel accedent" sign. [18:25] :-) [19:00] someone alert MC Hammer, it's OCR time! [19:07] that didn't take long [20:05] chrs [20:16] chr done [20:17] had a paying customer complaining about openid and private email list archives https://bugs.launchpad.net/launchpad/+bug/663923 [20:17] <_mup_> Bug #663923: Cannot view list archive of private team

< https://launchpad.net/bugs/663923 > [20:26] gary_poster: have you stocked up on milk and D batteries? [20:27] bac, milk, yes, D batteries, not so sure, but hopefully. That remonds me, we were going to plug the iPad in to make sure it had juice to amuse the kiddies in the dark. :-) how about you, bac? [20:27] gillam's parents and brother from nags head showed up yesterday with three coolers of food [20:28] so we won't starve [20:28] heh [20:28] cool [20:28] might go crazy and kill some people, though [20:28] stay away from us when you feel that way, then! [20:28] :) [20:28] jojo is very stressed and won't leave my side [20:31] oh, benji, bac, I'm going to go to the clojure conj (http://clojure-conj.org/) in Raleigh, using my conference days. I'm seeing if I can use training money to defray the registration cost--I need to check with Sarah about that. ClojureScript (Clojure plus Google Closure) is a potentially big enough deal that it is moderately arguable for me to spend the time, and I just wanted to anyway, and Francis was ok with i [20:31] that was intended as an invitation ;-) [20:31] benji, I expect you need to be home [20:31] but if you wanted to stay here for the conference, you certainly could [20:31] bac, poor jojo [20:32] brb [20:32] gary_poster: i'll be away those days in california [20:32] oh right [20:33] gary_poster: I'd love to go. I somewhat doubt I can swing being away from home though. Let me think about it and I'll get back to you on Monday. [20:39] cool benji [20:57] y'all have a nice weekend. stay dry. [21:01] bac: same to you