[04:36] lifeless, i'm going to have a stab at ssh reconnection [04:37] poolie: awesome! [04:41] will be awesome if i fix it, anyhow [04:43] poolie: its awesome that you're looking at it, really. [04:43] poolie: much appreciated === AuroraBorealis is now known as aurora|afk [04:59] Hi poolie - quick Q: There was some discussion here last Friday regarding the new config file features, and I'm not sure if that changes the next step for my config patch or not. I got the impression that maybe the PatchMatching idea was being dropped? [04:59] i don't actually know what that is [05:01] oh - hmm, it seems the devnotes have changed since then anyway :) [06:54] poolie: ping [06:55] hi stylesen [06:55] poolie: pm ? === aurora|afk is now known as AuroraBorealis === Mkaysi_ is now known as Mkaysi [09:28] i've pulled one branch from LP, converted it to 2a format and now pushing back to my account at LP...is starts speedy, but then it slows down..as a hell...yesterday i had to kill it after 6 hours...it's ~3500revs...any idea what might be wrong? [09:28] you haven't converted the copy on LP [09:28] at a guess [09:29] i've converted locally [09:29] are you pushing to the same branch on LP ? [09:30] the original branch is lp:~mdipierro/web2py/devel and i'm pushing to lp:~gour/web2py/devel [09:31] here is what i see now: 105548kB 28kB/s / Fetching revisions:Inserting stream:Estimate 51443/54666 and after it arrives at 51000 it crawls [09:32] interesting [09:32] uhm [09:32] 3500revs is not so big, imho [09:32] I suggest filing a question (or asking here in an hour or so - I have to go, and I don't think jam or jelmer or vila are here yet) [09:33] lifeless: ok. thank you...i'll wait here === gour1 is now known as gour [12:25] jelmer: hello. lifeless told me to ask you, vila or jam about my problem of slow upload to LP...do you mind? [12:26] hi gour [12:27] gour: what's slow exactly? [12:27] jelmer: push...after some time it starts crawling [12:28] jelmer: the original branch is lp:~mdipierro/web2py/devel and i'm pushing to lp:~gour/web2py/devel after i converted branch to 2a format [12:28] today i got 'Write failed: Broken pipehing revisions:Inserting stream:Estimate 51485/54666' and yesterday i killed push after not being complete in 6 hours [12:29] my adsl is 256k/4M [12:30] gour: how big is the repository? [12:30] jelmer: ~3500revs...let me size it [12:32] jelmer: hmm...tar.gz is 1.1G...no wonder, but i see lot of stuff like 'web2py/.bzr/repository/obsolete_packs/...' is it normal? [12:33] let me check what is the size of original branch in old format [12:33] the obsolete packs can be removed [12:34] just remove obsolete_packs directory? [12:35] there's no need to remove it, but you should be able to, and please exclude it when looking at the size of the repo [12:36] i'm pulling orignal repo now [12:39] obsolete_packs occupy 590M and ../repository/packs is 577M [12:55] Don't remove the _directory_. Removing the files _in_ it should be fine. [13:00] jelmer: here http://pastebin.com/pQ9LQvVA is du for old format and here http://pastebin.com/bk3xFdG6 for a 2a format which is 2x bigger...is it normal? [13:01] gour: that seems to be purely because of the obsolete packs [13:01] so it's normal, as the obsolete packs can be removed [13:01] why are they created? [13:02] Because some filesystems are lying liars that lie. [13:03] (and even when they aren't, the hardware picks up the slack by lying on their behalf) [13:03] heh [13:04] I rm-ed obsolete packs...now checking repo [13:05] gour: lp:~gour/web2py/devel doesn't appear to exist [13:05] jelmer: yes, it did not complete [13:05] so i rm-ed it [13:06] gour: I'm wondering if perhaps it was in an older format, and bzr was trying to convert back from 2a into that format [13:06] while doing the push [13:06] lp:~gour/web2py/devel is supposed to be the original branch in 2a format [13:06] jelmer: no, it was not exuisting before [13:07] however, original web2py is still in old format [13:08] now 2a branch repo is 591M...let me try to push it [13:11] gour: not having the obsolete packs won't impact performance in any way [13:12] What might LP's stacking stuff do, to pushing a 2a variant of an older format branch? [13:16] let me see if it will complete now in reasonable time... [13:27] it arrived at ~4850/51093...and now it is becoming very slow... [14:30] i cloned original hg repo from google and pushed it to bitbucket...it did the job in ~15mins (14:09,85 total) while pushing to LP is still crawling...it's very frustrating [14:40] gour: ah, the original is an import from hg? [14:40] gour: that might be related [14:40] jelmer: yes [14:41] but don't know how it's done [14:41] what was the original again? [14:41] lp:~mdipierro/web2py/devel [14:42] i'm resisting to use git (bzr-git is great), but there are more & more projects using hg...and things like this make it cumbersome [14:45] hmm, doesn't appear to've be done with bzr-hg [14:46] * jelmer tries reproducing the issue [14:46] jelmer: original repo is at hg clone https://code.google.com/p/web2py/ [14:46] jelmer: however, bzr branch is regularly synced [14:47] gour: not using bzr-hg, that doesn't support the pack-0.92 format [14:48] well, i do not use bzr-hg at all for this repo [14:52] * gour quit-ed push to LP... [14:53] and i wonder how he converts from hg to bzr... [14:55] gour: I'm running the upgrade here now, will try to push in a second [14:56] jelmer: thanks a lot [15:03] jelmer: I'm pretty surprised about how many web2py branches are on Launchpad, given that there isn't a development focus, so they should all be 500MB or so === jam1 is now known as jam [15:05] anyway, time to pick up my son, catch you guys later [15:05] hdy jzm [15:05] I mean [15:05] Hey jam :) [15:05] jam: Have a nice evening [15:05] Shoot, you missed by the _whole_ alphabet! :p [15:05] lol [15:06] * jelmer hits fullermd on the head with the alphabet [15:06] * fullermd is consonantly abused like this :( [15:09] hehe [15:21] jelmer: i've to go out. bbl in ~2hrs [15:22] gour: k [15:52] gour: push just finished here: bzr push lp:~jelmer/web2py/test 10.66s user 1.19s system 0% cpu 42:35.45 total === abentley is now known as abentley-lunch [16:14] gour: it looks like the hg repository has significantly different contents [16:19] jelmer: I think it was some time ago and maxb even told us about it but I forgot to say thanks for fixing bzr in LP so we have KDE imports working [16:19] Quintasan: np :) [16:19] * Quintasan is sure forgetful [16:23] Quintasan: does that mean all recipes are working now? [16:23] jelmer: If the code inside the branches is working then yes :P [16:27] :) === beuno is now known as beuno-lunch [16:37] jelmer: I got a notification that a fix is ready regarding the bug about lp:gcc svn import timeout [16:37] Is it to not fetch the tags at all? [16:38] jimis: it will fetch the tags but not the revisions those tags are referencing, if those revisions are not part of the ancestry of the branch [16:38] jimis: this is consistent with what it did earlier [16:47] jelmer: did you try increasing the timeout value in Twisted? [16:47] I haven't worked with Twisted, but here is a relevant thread I found, in case you missed it: [16:48] http://osdir.com/ml/python.twisted.web/2006-02/msg00048.html === kedare is now known as kedare_ifr === abentley-lunch is now known as abentley [17:07] jimis: I'm not sure I follow; how is twisted relevant here? [17:11] jelmer: what do you mean 'different contents' ? [17:12] jelmer: http://launchpadlibrarian.net/73597623/vcs-imports-gcc-trunk.log [17:12] you have a nice conenction considering it took 42mins only [17:13] jelmer: Isn't it the same error, twisted.internet.error.TimeoutError? [17:21] gour: which bzr version you have? [17:21] jimis: 2.3.4 [17:22] I've seen major improvements with 2.4, related to how much data is actually downloaded [17:22] I wouldn't know if that's relevant to your case however [17:23] lp:gcc is a huge branch, and it's *unworkable* with 2.3 [17:23] jimis: that would be nice improvment..the present sitaution is almost like showstopper [17:23] 2.4 is working just fine though, even if it feels a bit sluggish (expected with huge repos like this) [17:24] gour: I'm not sure if it will help your case, but it's easy to try the beta [17:24] jimis: i may try it [17:25] just make sure your python is >= 2.6 [17:25] 2.7.2 here [17:27] gour: there is a different number of revisions in both repositories, and the actual files seem to differ [17:27] gour: I think my bzr version probably matters as well - what are you running? [17:27] hmm...strange [17:27] jelmer: 2.3.4 [17:27] gour: So I'm curious how one branch is created from the other [17:27] me too [17:28] jimis: the error message is the same, but it's got nothing to do with the actual problem [17:28] jelmer: ok, I wouldn't know [17:28] I saw a timeout, so I thought increasing the socket timeout would mitigate the error... [17:29] Latency may contribute too; you're probably closer to LP than he is. [17:30] Though one would hope that the smarts are smart enough that that's unimportant in such a case... [17:30] I'm running 2.4b5, which I suspect will matter. [17:30] i see [17:33] gour: the bzr repository includes a bunch of tar files, the hg one does not [17:33] Just my opinion here: Guys, try to push 2.4 forward as fast as possible, older versions make bzr look like a fool for big projects... [17:34] I had lost several hours waiting on 2.3 to finish until I came to this channel and you told me to try 2.4 beta [17:34] jimis: 2.4.0 will be released in 2 days (: [17:34] good :-) [17:34] gour: the hg repo history doesn't start until dec 2009, until well after those tarballs were removed from the bzr repo [17:35] gour: another thing that will help is setting the development focus for web2py on launchpad [17:35] gour: and upgrading the main branch to 2a [17:35] that will mean every push of a new branch to launchpad won't push up any revisions that are in the development focus [17:36] jimis: I agree, although it only matters for the *really* big projects like gcc [17:39] jelmer: 2.4-final will be incompatible with python 2.4 after all? [17:40] since I'm running it on RHEL I had to fight around this issue too === beuno-lunch is now known as beuno [17:40] it's fine now, I'm just curious [17:42] i'm pondering ways to bring the benefits to revision control to people in my company that spend most of their time in excel and email. Is there a really, really simple, graphical interface to bzr that I should consider? [17:56] indigo: qbzr/bzr-explorer/tortoisebzr are the ones to consider I think [17:57] I don't have much experience using a graphical interface exclusively [17:57] jimis: yeah [18:03] jelmer: i suggested main dev to upgrade to 2a, but wanted to try it ourself first...and those problems are not really auspicious to push migration... [18:04] is there anything special in usinb bzr in-place? i'd like to try 2.4beta [18:08] gour: using it in place should be fine, except for plugins that might give warnings [18:08] jelmer: ok, i did 'home' install [18:14] * gour is pushing with 2.4b5 [18:14] gour: you probably want to make sure the extensions were built and installed too [18:15] jelmer: i believe that, based on --version output, it's ok [18:17] hmm...warning about some extensions [18:18] but launchpad is 2.4.b5 [18:18] gour: I mean the C extensions [18:19] jelmer: that was built [20:33] Hey, all. Using Ubuntu 11.04, with Emacs Snapshot GTK, vc-bzr, and have gpg-agent installed. I'm still having trouble signing my commits. I get the following error: http://paste.lisp.org/display/123944 [20:33] I've checked (both via shell-mode and getenv) that my GPG_AGENT_INFO is available to emacs. [20:34] Google is not much my friend in this regard, but I've searched for 'vc-bzr gpg sign commits' and the like [21:37] urbanape: does it work outside of vc-bzr? === Quintasan_ is now known as Quintasan [23:52] hi all [23:52] hai [23:58] jelmer: yes, it does, even from within emacs (shell-mode)