2011/08/10/#bzr.txt

[04:36] <poolie> lifeless, i'm going to have a stab at ssh reconnection
[04:37] <lifeless> poolie: awesome!
[04:41] <poolie> will be awesome if i fix it, anyhow
[04:43] <lifeless> poolie: its awesome that you're looking at it, really.
[04:43] <lifeless> poolie: much appreciated
=== AuroraBorealis is now known as aurora|afk
[04:59] <thomi> 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] <poolie> i don't actually know what that is
[05:01] <thomi> oh - hmm, it seems the devnotes have changed since then anyway :)
[06:54] <stylesen> poolie: ping
[06:55] <poolie>  hi stylesen
[06:55] <stylesen> poolie: pm ?
=== aurora|afk is now known as AuroraBorealis
=== Mkaysi_ is now known as Mkaysi
[09:28] <gour> 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] <lifeless> you haven't converted the copy on LP
[09:28] <lifeless> at a guess
[09:29] <gour> i've converted locally
[09:29] <lifeless> are you pushing to the same branch on LP ?
[09:30] <gour> the original branch is lp:~mdipierro/web2py/devel and i'm pushing to  lp:~gour/web2py/devel
[09:31] <gour> 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] <lifeless> interesting
[09:32] <lifeless> uhm
[09:32] <gour> 3500revs is not so big, imho
[09:32] <lifeless> 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] <gour> lifeless: ok. thank you...i'll wait here
=== gour1 is now known as gour
[12:25] <gour> jelmer: hello. lifeless told me to ask you, vila or jam about my problem of slow upload to LP...do you mind?
[12:26] <jelmer> hi gour
[12:27] <jelmer> gour: what's slow exactly?
[12:27] <gour> jelmer: push...after some time it starts crawling
[12:28] <gour> 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] <gour> 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] <gour> my adsl is 256k/4M
[12:30] <jelmer> gour: how big is the repository?
[12:30] <gour> jelmer: ~3500revs...let me size it
[12:32] <gour> 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] <gour> let me check what is the size of original branch in old format
[12:33] <jelmer> the obsolete packs can be removed
[12:34] <gour> just remove obsolete_packs directory?
[12:35] <jelmer> 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] <gour> i'm pulling orignal repo now
[12:39] <gour> obsolete_packs occupy 590M and ../repository/packs is 577M
[12:55] <fullermd> Don't remove the _directory_.  Removing the files _in_ it should be fine.
[13:00] <gour> 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] <jelmer> gour: that seems to be purely because of the obsolete packs
[13:01] <jelmer> so it's normal, as the obsolete packs can be removed
[13:01] <gour> why are they created?
[13:02] <fullermd> Because some filesystems are lying liars that lie.
[13:03] <fullermd> (and even when they aren't, the hardware picks up the slack by lying on their behalf)
[13:03] <gour> heh
[13:04] <gour> I rm-ed obsolete packs...now checking repo
[13:05] <jelmer> gour: lp:~gour/web2py/devel doesn't appear to exist
[13:05] <gour> jelmer: yes, it did not complete
[13:05] <gour> so i rm-ed it
[13:06] <jelmer> 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] <jelmer> while doing the push
[13:06] <gour> lp:~gour/web2py/devel is supposed to be the original branch in 2a format
[13:06] <gour> jelmer: no, it was not exuisting before
[13:07] <gour> however, original web2py is still in old format
[13:08] <gour> now 2a branch repo is 591M...let me try to push it
[13:11] <jelmer> gour: not having the obsolete packs won't impact performance in any way
[13:12] <fullermd> What might LP's stacking stuff do, to pushing a 2a variant of an older format branch?
[13:16] <gour> let me see if it will complete now in reasonable time...
[13:27] <gour> it arrived at ~4850/51093...and now it is becoming very slow...
[14:30] <gour> 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] <jelmer> gour: ah, the original is an import from hg?
[14:40] <jelmer> gour: that might be related
[14:40] <gour> jelmer: yes
[14:41] <gour> but don't know how it's done
[14:41] <jelmer> what was the original again?
[14:41] <gour> lp:~mdipierro/web2py/devel
[14:42] <gour> 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] <jelmer> hmm, doesn't appear to've be done with bzr-hg
[14:46]  * jelmer tries reproducing the issue
[14:46] <gour> jelmer: original repo is at hg clone https://code.google.com/p/web2py/
[14:46] <gour> jelmer: however, bzr branch is regularly synced
[14:47] <jelmer> gour: not using bzr-hg, that doesn't support the pack-0.92 format
[14:48] <gour> well, i do not use bzr-hg at all for this repo
[14:52]  * gour quit-ed push to LP...
[14:53] <gour> and i wonder how he converts from hg to bzr...
[14:55] <jelmer> gour: I'm running the upgrade here now, will try to push in a second
[14:56] <gour> jelmer: thanks a lot
[15:03] <jam1> 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] <jam> anyway, time to pick up my son, catch you guys later
[15:05] <jelmer> hdy jzm
[15:05] <jelmer> I mean
[15:05] <jelmer> Hey jam :)
[15:05] <jelmer> jam: Have a nice evening
[15:05] <fullermd> Shoot, you missed by the _whole_ alphabet!   :p
[15:05] <gour> lol
[15:06]  * jelmer hits fullermd on the head with the alphabet
[15:06]  * fullermd is consonantly abused like this   :(
[15:09] <jelmer> hehe
[15:21] <gour> jelmer: i've to go out. bbl in ~2hrs
[15:22] <jelmer> gour: k
[15:52] <jelmer> 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] <jelmer> gour: it looks like the hg repository has significantly different contents
[16:19] <Quintasan> 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] <jelmer> Quintasan: np :)
[16:19]  * Quintasan is sure forgetful
[16:23] <jelmer> Quintasan: does that mean all recipes are working now?
[16:23] <Quintasan> jelmer: If the code inside the branches is working then yes :P
[16:27] <jelmer> :)
=== beuno is now known as beuno-lunch
[16:37] <jimis> jelmer: I got a notification that a fix is ready regarding the bug about lp:gcc svn import timeout
[16:37] <jimis> Is it to not fetch the tags at all?
[16:38] <jelmer> 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] <jelmer> jimis: this is consistent with what it did earlier
[16:47] <jimis> jelmer: did you try increasing the timeout value in Twisted?
[16:47] <jimis> I haven't worked with Twisted, but here is a relevant thread I found, in case you missed it:
[16:48] <jimis> 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] <jelmer> jimis: I'm not sure I follow; how is twisted relevant here?
[17:11] <gour> jelmer: what do you mean 'different contents' ?
[17:12] <jimis> jelmer: http://launchpadlibrarian.net/73597623/vcs-imports-gcc-trunk.log
[17:12] <gour> you have a nice conenction considering it took 42mins only
[17:13] <jimis> jelmer: Isn't it the same error, twisted.internet.error.TimeoutError?
[17:21] <jimis> gour: which bzr version you have?
[17:21] <gour> jimis: 2.3.4
[17:22] <jimis> I've seen major improvements with 2.4, related to how much data is actually downloaded
[17:22] <jimis> I wouldn't know if that's relevant to your case however
[17:23] <jimis> lp:gcc is a huge branch, and it's *unworkable* with 2.3
[17:23] <gour> jimis: that would be nice improvment..the present sitaution is almost like showstopper
[17:23] <jimis> 2.4 is working just fine though, even if it feels a bit sluggish (expected with huge repos like this)
[17:24] <jimis> gour: I'm not sure if it will help your case, but it's easy to try the beta
[17:24] <gour> jimis: i may try it
[17:25] <jimis> just make sure your python is >= 2.6
[17:25] <gour> 2.7.2 here
[17:27] <jelmer> gour: there is a different number of revisions in both repositories, and the actual files seem to differ
[17:27] <jelmer> gour: I think my bzr version probably matters as well - what are you running?
[17:27] <gour> hmm...strange
[17:27] <gour> jelmer: 2.3.4
[17:27] <jelmer> gour: So I'm curious how one branch is created from the other
[17:27] <gour> me too
[17:28] <jelmer> jimis: the error message is the same, but it's got nothing to do with the actual problem
[17:28] <jimis> jelmer: ok, I wouldn't know
[17:28] <jimis> I saw a timeout, so I thought increasing the socket timeout would mitigate the error...
[17:29] <fullermd> Latency may contribute too; you're probably closer to LP than he is.
[17:30] <fullermd> Though one would hope that the smarts are smart enough that that's unimportant in such a case...
[17:30] <jelmer> I'm running 2.4b5, which I suspect will matter.
[17:30] <gour> i see
[17:33] <jelmer> gour: the bzr repository includes a bunch of tar files, the hg one does not
[17:33] <jimis> 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] <jimis> 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] <jelmer> jimis: 2.4.0 will be released in 2 days (:
[17:34] <jimis> good :-)
[17:34] <jelmer> gour: the hg repo history doesn't start until dec 2009, until well after those tarballs were removed from the bzr repo
[17:35] <jelmer> gour: another thing that will help is setting the development focus for web2py on launchpad
[17:35] <jelmer> gour: and upgrading the main branch to 2a
[17:35] <jelmer> 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] <jelmer> jimis: I agree, although it only matters for the *really* big projects like gcc
[17:39] <jimis> jelmer: 2.4-final will be incompatible with python 2.4 after all?
[17:40] <jimis> since I'm running it on RHEL I had to fight around this issue too
=== beuno-lunch is now known as beuno
[17:40] <jimis> it's fine now, I'm just curious
[17:42] <indigo> 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] <jelmer> indigo: qbzr/bzr-explorer/tortoisebzr are the ones to consider I think
[17:57] <jelmer> I don't have much experience using a graphical interface exclusively
[17:57] <jelmer> jimis: yeah
[18:03] <gour> 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] <gour> is there anything special in usinb bzr in-place? i'd like to try 2.4beta
[18:08] <jelmer> gour: using it in place should be fine, except for plugins that might give warnings
[18:08] <gour> jelmer: ok, i did 'home' install
[18:14]  * gour is pushing with 2.4b5
[18:14] <jelmer> gour: you probably want to make sure the extensions were built and installed too
[18:15] <gour> jelmer: i believe that, based on --version output, it's ok
[18:17] <gour> hmm...warning about some extensions
[18:18] <gour> but launchpad is 2.4.b5
[18:18] <jelmer> gour: I mean the C extensions
[18:19] <gour> jelmer: that was built
[20:33] <urbanape> 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] <urbanape> I've checked (both via shell-mode and getenv) that my GPG_AGENT_INFO is available to emacs.
[20:34] <urbanape> Google is not much my friend in this regard, but I've searched for 'vc-bzr gpg sign commits' and the like
[21:37] <jelmer> urbanape: does it work outside of vc-bzr?
=== Quintasan_ is now known as Quintasan
[23:52] <poolie> hi all
[23:52] <lifeless> hai
[23:58] <urbanape> jelmer: yes, it does, even from within emacs (shell-mode)