[01:11] <bobweaver> Hello there I am using bzr explorer and I was going to push code and I entered the wrong password (caps was on)  but I can not fix this error . I am sure that it is right under my nose but I can not seem to fix this. I tried to sign in from command line to launchpad and all is well there
[01:11] <poolie> hi there
[01:12] <bobweaver> Hi poolie
[01:12] <poolie> you can't fix the problem meaning that it's not giving you a second chance to enter the password?
[01:12] <poolie> that's a bit icky
[01:12] <poolie> you should be able to find and edit authentication.conf
[01:12] <poolie> i think it's stored there
[01:12] <bobweaver> yup ^^ no 2nd chance
[01:12] <bobweaver> ohh cool thanks
[01:12] <poolie> you could file a bug at pad.lv/fb/bzr-explorer
[01:12] <bobweaver> cool will do
[01:12] <bobweaver> thanks poolie  !
[01:13] <poolie> np
[02:02] <bobweaver> gezz I still can get it worked out under that file there is no passwd this is the error that I am getting bzr: ERROR: Could not acquire lock "(remote lock)":       I have tried to also break the lock but that also did not do anything
[02:08] <poolie> bobweaver, sometimes that means you don't have permission on the remote machine?
[02:11] <bobweaver> my ssh key is there
[02:15] <bobweaver> I am going to try and reboot
[02:21] <bobweaver> so a no go let me enter password and I def typed correct this time. must be lp ?
[02:22] <bobweaver> I can log into LP and all my old code is there and asys that it is owned by me
[02:23] <kbulgrien> so if I have a shared repo somewhere on a machine that isn't up all the time, does it make any sense to rsync it over to another machine, or is that messed up thinking?
[02:23] <kbulgrien> IE. what might be wierd about having multiple shared repos around.
[02:25] <kbulgrien> I imagine I should have set up the shared repo on a server... and I guess I can dismantle the other one... but then might there be some benefit to not dismantling the original.
[02:31] <fullermd> kbulgrien: I can't think of anything particular to shared repos that makes it any different from doing the same thing with any other set of branches.
[02:31] <fullermd> kbulgrien: It just sounds like a management question of making sure people are pushing things into the right place, not somewhere that'll get overwritten.
[02:34] <kbulgrien> can you sync the shared repos with bzr, or is just rsync the way to do that?
[02:36] <fullermd> There's nothing in bzr at the moment that syncs repos.  There's some stuff like multi-pull in bzrtools that gives you some automation of iterating over the branches, but it's all branch-at-a-time.
[02:36] <fullermd> rsync is simpler.  Could potentially lose you stuff if the right (wrong) things blow up in the middle.
[02:38] <kbulgrien> I suppose --delete is  out of the question.  files never go away in the repo?
[02:38] <kbulgrien> only add?
[02:39] <kbulgrien> Along the lines that you could do two rsync one in each direction to make them the same.
[02:39] <kbulgrien> if you were to have had things put separately in each.
[02:39] <fullermd> Files go away with fair regularity.
[02:40] <kbulgrien> ok
[02:40] <fullermd> Multiple old files get repacked into one new file.  With bad timing, you could get the old files --delete'd before the new one gets synced, and then the squirrel takes the last bite through the network cable...
[02:41] <kbulgrien> :-)
[02:41] <lifeless> nom nom nom
[02:41] <fullermd> To be sure, it adds that extra savor to the squirrel stew that evening, but you get yelled at a lot before then.
[02:42] <kbulgrien> its always faster doing it the second time eh? ;-)
[02:46] <kbulgrien> maybe --delay-updates would foil the squirrel
[02:51] <kbulgrien> Probably mostly it would make sense for one copy just to be the big giant backup that nobody ever uses except the mirror or the owner in a recovery situation.
[02:51] <fullermd> Yeah.  You definitely wouldn't want to try using rsync as a "keep both in sync while people are writing to both" thing.
[02:52] <mgrandi> seems like just a simple script to run 'bzr update' occasionally on whatever branch
[02:52] <fullermd> You _could_ use bzr itself for that (I mean, that's just distributed development, right?) but I wouldn't try it automated.  So it sounds like just extra complexity in that case too.
[02:52] <mgrandi> seems like the best way
[02:52] <mgrandi> cause i assume it knows how to deal with whats happening if its being updated at the same point you are trying to download it
[02:53] <lifeless> so the problem is
[02:53] <mgrandi> and i think that the bzr docs say thats the preferred method
[02:53] <lifeless> rsync isn't file-system order
[02:53] <lifeless> bzr's primitives ensure ~0 data loss window if they are replayed in filesystem order
[02:53] <fullermd> Obviously, the solution is "zfs send"  8-}
[02:54] <lifeless> an external actor (like rsync) doing any of breadth-first, depth-first, etc traversals, with immediate or deferred deletes will create otherwise impossible situations in the receiving copy, if they happen concurrently with bzr acting on the repository
[02:55] <lifeless> this is the same reason that rsyncing a postgresql DB has no guarantees in the absence of cooperation-with-postgresql
[02:55] <lifeless> and we haven't written a cooperate-with-rsync/cp/etc module for bzr yet
[02:55] <lifeless> (FWIW git and hg have the same issue, its not unique to bzr)
[02:55] <mgrandi> yeah.
[02:55]  * kbulgrien looks for a repo dump
[02:55] <mgrandi> thats wht the docs just suggest
[02:55] <mgrandi> using bzr update/pull
[02:56] <mgrandi> cause then that way it handles things correctly
[02:56] <lifeless> one of the things that makes doing a cooperative thing for bzr is that we assume that fsync doesn't work ;) [for good reason]
[03:01] <kbulgrien> ok multi-pull seems agnostic to what branches exist in a particular folder, so it could just pull everything it finds without someone having to say what to pull.
[03:01] <mgrandi> i believe it pulls all branches in a shared repo.
[03:01] <fullermd> No, it doesn't care about repos one way or another.  It's roughly equivalent to `find . -type bzr-branch -print | xargs -n1 bzr pull`
[03:03] <kbulgrien> hmm and there is automirror
[03:03] <fullermd> So, it'll pull a set of independent branches, the branches in one repo, the branches in a couple repos, some combination...  whatever it finds under the dir you point it at.
[03:03] <kbulgrien> its kind of a multipush I guess
[03:04] <kbulgrien> ok, cool, learned something... I found the backup/restore stuff in the admin guide.
[03:06] <kbulgrien> I think I read that if, I am the only one using stuff, rsync is faster and ok.  Its when you can't guarantee the copied stuff is static that that becomes an issue.
[03:07] <kbulgrien> That's what I gathered from the above comments too, so I think I get it.
[03:07] <lifeless> rsync *can* be faster; it won't right after compactions
[03:08] <kbulgrien> I would probably be better off sticking to bzr stuff though for sake of making myself more well-rounded and familiar with the tool set.
[03:09] <kbulgrien> not to mention forgetting the nitpick and messing myself over once I was so used to using the unsafe tool.
[03:10] <kbulgrien> I'll decline to admit how long I rsync backup'd postgres in the middle of the night ;-) before I wrote a dump script.
[03:11] <kbulgrien> all the time knowing it wasn't really safe.
[03:11] <fullermd> I do it every night.
[03:12] <fullermd> Granted, I mostly care about and rely on the gzip'd dump file also in the dir tree being synced, but..
[03:14] <kbulgrien> I actually have mine pulling over dumps and the receiving system drops and rebuilds everything based on what it just pulled over.
[03:15] <kbulgrien> Anyway... trying to get geared up to do bzr stuff in a way so I have a hot backup.
[05:43] <grettke> Hi folks. I would like to check out files both on windows and linux and have bzr ignore differing line endings. My rules has a name * eol = native but that isn't helping. What am I doin wrong here?
[05:43] <poolie> hm, that should do it
[05:43] <mgrandi> does bzr try to convert line endings for you?
[05:43] <mgrandi> that seems weird o.o
[05:44] <poolie> what do you mean more specifically by 'not helping'?
[05:44] <grettke> poolie: when I run a bzr st, all files are reported as changed, but when I do a qdiff and "Ignore whitespace differences" there are no changes
[05:45] <grettke> mgrandi: I don't know but if it were then that would be a whitespace change in every file wouldn't it...
[05:46] <mgrandi> i dunno, i thought bzr would just bersion the file and not try and modify line endings
[05:49] <fullermd> If you hit 'commit', does it actually try to commit those changes, or does it say there's nothing to do?
[05:50] <grettke> I didn't try committing it. Odd thing is that when I do a revert... it doesn't revert it still reports modifications.
[05:57] <spiv> mgrandi: only if you configure it to do so
[05:58] <mgrandi> ah
[08:05] <mgz> morning
[08:06] <mgrandi> hey
[08:06] <mgrandi> if you are up that means i'm up way too late. hmm
[08:07] <fullermd> I think it just means he's up way too early.
[08:07] <fullermd> I mean, the sun probably isn't even down yet where he is.
[08:07] <mgrandi> so, i did a revert the other day and i cant revert the tree cause its says TreeTransform is malformed
[08:07] <mgrandi> wat do
[08:13] <mgrandi> or rather should i just report a bug / upload the branch
[08:16] <mgz> so, the problem with TreeTransform errors, is without a clear way to reproduce there's not much to go on
[08:16] <mgz> ideally you should work out what change is problematic, and reproduce it in a fresh branch, giving steps in bug report
[08:17] <mgz> it's generally easy enough to work around not being able to revert something, just branch to a new location from the rev that you're trying to revert from
[08:18] <mgz> a bug report and packed up branch wouldn't hurt though, if you confirm that someone downloading and trying the steps you give will get the same error
[08:19] <mgrandi> yeah, i zipped up the branch
[08:23] <mgrandi> and just report a bug then?
[08:24] <mgz> sure, but trying to reproduce from scratch woul still be very useful.
[08:24] <mgrandi> well i just tried it (with the branch that i zipped) and if you do bzr revert on it then it still crashes
[08:25] <mgrandi> how do you mean do it from scratch, as in make a new branch and try do the steps to break it?
[08:28] <mgz> right, and you can search the existing bugs to see if one of those already covers your case
[08:29] <mgrandi> k
[08:30] <mgz> eg bug 632704 and bug 660125
[08:30] <ubot5> Launchpad bug 632704 in Bazaar "Can't merge a directory deletion in a wt with a pending subdirectory deletion" [High,Confirmed] https://launchpad.net/bugs/632704
[08:30] <ubot5> Launchpad bug 660125 in Bazaar "shelve crashes with "Tree transform is malformed" when renamed file already exists" [Medium,Confirmed] https://launchpad.net/bugs/660125
[08:30] <mgz> working out what the particular case that's making it fail for you is the most relevent thing
[08:32] <mgrandi> i know i was doing some renaming stuff
[08:33] <mgrandi> but its late and i'll do it this weekend, and if jelmer is here i'm also having problems with bzr-git >.>
[08:34] <fullermd> What's 'stat' look like?
[08:34] <mgrandi> bzr status on the branch?
[08:34] <fullermd> Yah.  Well, on the WT, but 6.5 of one...
[08:36] <mgrandi> its ugly D: http://bpaste.net/show/26999/
[08:37] <mgrandi> not sure what the * means in 'modified'
[08:37] <fullermd> +/-x
[08:38] <fullermd> Hm.  How about this part.
[08:38] <mgrandi> ah, yeah, i never modified the +x by myself
[08:38] <fullermd> removed: php_frontend/src/codeigniter/system/libraries/Twig/lib/
[08:38] <fullermd> unknown:   php_frontend/src/codeigniter/system/libraries/Twig/lib/
[08:38] <fullermd> What if you rm (or move aside) that unknown dir?
[08:39] <mgz> that looks like a winner.
[08:39] <mgrandi> yeah
[08:39] <mgrandi> that fixed it.
[08:40] <mgrandi> so now the question is how it got into that weird state o.o
[08:44] <fullermd> A little fiddling with trivial similar setups doesn't show up any problems, so it's liable to be a little twisted.
[08:47] <mgrandi> i have the log so ill be able to recreate exactly what i did
[08:51] <jelmer> mgrandi: hey
[08:51] <mgrandi> hey
[08:52] <mgrandi> so, i have a bzr branch that i was using to do stuff with github
[08:52] <mgrandi> and one branch is just a straight up copy of someones github repo, i haven't done anything to it
[08:52] <mgrandi> i tried updating it and bzr-git says the branches have diverged, and to do 'bzr missing' to see why, but then bzr missing says 'git smart protocol doesn't support that'
[08:54] <jelmer> mgrandi: yeah, the git protocol doesn't support remote graph inspecition
[08:54] <mgrandi> then i just tried branching it again in a brand new branch, and i get: bzr: ERROR: Could not determine revno for {git-v1:78d796d1f018bb66d986e73bc4641b6c0b372a55} because its ancestry shows a ghost at {git-v1:78d796d1f018bb66d986e73bc4641b6c0b372a55}
[08:55] <jelmer> what version of bzr-git?
[08:55] <mgrandi> how do you get versions of the plugins?
[08:56] <jelmer> mgrandi: 'bzr plugins'
[08:56] <mgrandi> git 0.6.8dev
[08:56] <jelmer> hmmok, that's fairly recent
[08:56] <jelmer> I'm not sure what's going on there, without looking into it in more detail
[08:58] <mgrandi> it seems that its something to do with the repo, since making a folder outside of the repo i have works fine
[08:58] <jelmer> mgrandi: aha
[08:58] <jelmer> is the repo old, does it perhaps have revisions that might have been fetched with old (broken?) versions of bzr-git?
[08:59] <mgrandi> the repo isnt that old, and i think ive been using this version of bzr git for a while
[09:01] <mgrandi> the last commit i have in that branch is  Fri 2012-02-24 21:16:15 -0200
[09:01] <mgrandi> or that came from the guy whose github i'm branching
[09:03] <fullermd> Well, I give up.  I've tried a handful of likely-looking slices of those changes, and I don't get any blowups with 2.5 or .dev.
[09:04] <mgrandi> fullermd: i finally scrolled through my log so i have the exact commands i did
[09:04] <mgrandi> so i'll try and recreate it
[09:05] <fullermd> What version are you running?
[09:05] <mgrandi> 2.5.0
[09:06] <fullermd> Oh well.
[09:07] <mgz> the rename of files from a subdir of that clashing directory looks interesting for the revert
[09:07] <jelmer> mgrandi: so, I'm out of ideas too without looking into the issue more closely
[09:07] <mgrandi> do you want the entire shared repo / branch?
[09:08] <fullermd> http://paste.ubuntu.com/927613/   is the script I worked up for testing.
[09:08] <fullermd> But it works as expected for me, so it's something more than just moving files out of the subdir but leaving it sitting there.
[09:08] <fullermd> And more than having other stuff beside it also being moved around.
[09:09] <mgz> you're a twinkling object in the night sky, fullermd
[09:10] <mgrandi> im uploading the repo/branch/ and bzr.log output
[09:10] <fullermd> Yeah, I blow a lot of hot gas around sometimes.
[09:11] <mgrandi> its ugly cause im new to this framework and i couldn't figure out where to put this one library so thats why i have so many 'bzr move' and whatnot, but here it is:  www.kramidnarg.com/stuff/tree%20transform%20bug.zip
[09:42] <fullermd> Does that include the branch in the state where the revert blows up?
[09:42] <mgrandi> yeah
[09:42] <fullermd> Works for me.
[09:42] <mgrandi> in 'tmprepo.zip'
[09:42] <mgrandi> really? doing bzr revert inside tmprepo/enquest_work works for you?
[09:43] <fullermd> Yep.  With .dev and 2.5.0.
[09:43] <mgrandi> soooo why does mine not work then
[09:44] <fullermd> Now, here's one difference: You're apparently on OS X.
[09:45] <mgrandi> yeah
[09:45] <mgz> maybe a platform dif... what fullermd said
[09:45] <fullermd> Maybe the fs is doing something with the .DS_STORE's?
[09:45] <mgrandi> bzr 2.5.0 on python 2.6.7 (Darwin-11.3.0-x86_64-i386-64bit)
[09:46] <mgrandi> does bzr even use .ds_store? how would that effect it
[09:46] <mgrandi> and i think i have those ignored anyway
[09:46] <fullermd> OS X does do some special stuff with unicode normalization, but it looks like all the filenames are 7-bit ASCII, so it's probably not that.
[09:46] <fullermd> No, but maybe it hits some conflict with the forks when moving stuff around, since the OS treats them special?
[09:46] <mgz> my first guess was directory rename differences, but that was assuming windows rather than osx
[09:46]  * fullermd is guessing pretty randomly...
[09:47] <mgrandi> mac is not case sensitive...which is weird
[09:47] <mgrandi> maybe that?
[09:47] <mgz> allowing renaming over an existing (empty) dir is a posix thing, but osx should allow it too
[09:47] <fullermd> Mmm.  I thought it was.
[09:47] <mgz> case things is also a possibility.
[09:48] <mgrandi>  /users/markgrandi is the same as /Users/markgrandi, i dont know if python forces a case sensitivity
[09:48] <mgrandi> HFS+ supprts case sensitivity but mac os x doesn't do it for some reason
[09:48] <fullermd> Oh, right, it's case-insensitive-but-case-preserving.
[09:49] <fullermd> That's the most likely place, yah.  Nothing jumps out at me immediately though.
[09:49] <mgrandi> are you on windows or linux fullermd
[09:50] <fullermd> Neither, naturally.  I have [daemonic] standards  ;p
[09:50] <mgrandi> heh
[09:50] <mgrandi> i'll try on my windows/linux machines later
[09:50] <mgrandi> when its not 3 am
[09:52] <fullermd> Give http://paste.ubuntu.com/927663/ a try locally.
[09:52] <fullermd> (prob. have to fix the bzr path or something anyway)
[09:52] <fullermd> 's a few tweaks past the one I pasted earlier.  Still works for me, but since yours does too, maybe it'll blow up for you.
[09:52] <fullermd> Get you a step closer to an isolated testcase anyway.
[09:54] <mgrandi> dear ubuntu: why do i have to log in to download a paste. signed, me
[09:54] <fullermd> Anyway.  's way over my time quota; I've gotta get back to pretending to work in preparation for tomorrow's (later today's) pointless conference call.
[09:55] <mgrandi> ok
[09:55] <mgrandi> i'll let you know what happens
[10:33] <mgrandi> going to bed, i'll be on later to discuss these problems more, ttyl~
[22:19] <ccxCZ> how can I disable the spinner/speed on actions on ssh repositories?
[22:20] <ccxCZ> also can I make bzr use /bin/ssh instead of paramiko, so it reuses established master session?
[22:30] <ccxCZ> got it, BZR_SSH=openssh BZR_PROGRESS_BAR=none
=== yofel_ is now known as yofel