[00:20] I get this error when I branch bzr: bzr: ERROR: Revision {robertc@robertcollins.net-20050919044328-0205c679f3051340} not present in "thanks.txt-20050309044946-58141ea3091846d8". [00:20] It happens whether I use bzr.dev or bzr 1.3 [00:26] into a rich-root{,-pack} repository? [00:27] grrr, I just hit the rich root bug again as well :-( [00:37] bob2: into a standalone branch [00:38] My Scenario: [00:38] Reviewers have full checkouts of mainline to merge dev branchs into (update -> merge from dev1 -> review and commit). [00:38] QA has a lightweight checkout (update -> test and tag) [00:39] My Question: How do tags propagate in checkouts? [00:39] The QA tags show up for any new checkouts or brances made from mainline but they don't show up in reviewer checkouts that already have the tagged revision. [00:39] If a reviewer uses pull (which requires a path strangely) they do get the tags. Shouldn't update transfer tags like pull and shouldn't pull default, in a checkout, to the checked out location? [00:59] Good morning. [00:59] igc: good morning! I'm jealous :) [01:00] hi spiv [01:00] spiv: if it makes you feel better, it rained a lot :-) [01:05] dhello [01:09] poolie: good morning [01:10] hello spiv [01:11] back [01:16] hi lifeless [01:20] hi igc, welcome back [01:20] thanks === jamesh_ is now known as jamesh [01:30] jamesh: thanks for working on the set_revision_info hook stuff [01:31] igc: no problem. The API from your branch was better than mine anyway :) [01:32] igc: was there a particular thing you wanted the hook for? [01:32] (I wanted it for bzr-dbus) [01:33] jamesh: it was part of the work spiv and I started in the Chicago sprint to get good server-side hooks [01:36] hello [01:37] what is the proper procedure after merging branches that have diverged? [01:37] Stavros: fix any problems, test or read the diff, then commit [01:37] and everything will go in its proper place? [01:37] in the graph, i mean? [01:38] can you explain your question more? [01:38] well, i have two branches that have diverged and i merge on mine [01:38] and "bzr status" gives me some weird stuff, pending merges and the like [01:38] right :) [01:38] and i am just wondering if that is proper behaviour [01:39] that's correct [01:39] ah, good [01:39] now commit on your branch [01:39] so if i commit everything will work fine? [01:39] good [01:39] jamesh: in combination with the code I recently landed to add a Branch.set_last_revision_info smart verb, it starts making some server-side hooks possible [01:39] okay great, thanks! [01:39] now, depending on the purpose of the two branches you may want to take one further step [01:39] if they are both _meant_ to be the same, [01:39] what's that? [01:39] spiv: I suppose you don't see the pre/post_commit hooks on the server side? [01:40] for example if you are working on the same bug on your laptop and desktop [01:40] since it is just pushing a locally committed revision [01:40] you may want to push your merge from the branch where you committed it into the other [01:40] ah [01:40] New bug: #217031 in bzr ""bzr launchpad-login -v MYNAME" gives no feedback" [Undecided,New] https://launchpad.net/bugs/217031 [01:40] that will switch the other branch to be in the same revision- you can then push and pull between them without needing to merge [01:40] however [01:40] if the branches have different purposes [01:40] well i have a different workflow with a master server, but i see what you mean [01:40] jamesh: right [01:40] for example if one is the trunk and one is your feature branche [01:40] yes [01:40] that's it [01:41] then you shouldn't do this [01:41] cool [01:41] yes [01:41] glad i could help [01:41] so i just wait until i finish the feature then [01:41] thanks a lot! [01:41] jamesh: and for some cases at least, you might not care if it was a new commit vs. a push vs. uncommit or whatever, you might want to always run a hook if the branch's tip changed. [02:07] abentley: ping; you've been talking with alexander about line ending stuff; is he doing roughly what we planned in London, or something different? [02:07] I don't have a clear idea what he's doing. [02:07] Sorry, gotta go. [02:07] bye [02:08] I'll read the thread in detail then [02:23] poolie: 1.4 made [02:32] i am tagging revisions with already existing tags, but pushing with --overwrite so the tags will get accepted is a bit dangerous, isn't it? [02:44] martin pool, you be here? [02:44] hersonls: he's poolie [02:44] poolie, hey guy [02:44] poolie, you be here? [02:53] hersonls: I suggest you ask the question/start the discussion you want to have; 'presence' on IRC is a fairly weak concept. [02:55] lifeless, tanks [02:59] bzr internal error: http://pastebin.com/m373ccb16 [03:00] and is there no way to specify a path delimiter with bzr? e.g. git add -- foo/; git commit -- foo/ seems to have no analogue in bzr [03:00] as in make it non-recursive? [03:01] no, as in not have to specify each of the files in the foo directory to bzr commit individually [03:01] wildfire: bzr commit foo should do that [03:01] bzr commit -m 'blah foo/ will commit all changes to foo and below [03:02] hmm, apparently I've managed to break the repository and I get the internal error now; so I can't check that that works [03:02] even on 'bzr status' [03:03] hersonls: go ahead [03:03] poolie, hey, you remeber me? [03:04] yes [03:04] poolie, tanks for help [03:05] bob2, poolie: oh, it appears me doing that command 'bzr commit blosxom' is what caused the internal error [03:14] wildfire: is bloxsom versioned? [03:14] wildfire: does 'bzr st' work? [03:14] wildfire: and do you have nested trees? [03:16] poolie, how i crate new team in lauchpad? [03:17] lifeless, it was versions, bzr st fails with an internal exception (assuming it is the same, and assuming 'st' is 'status'), not a nested tree, no [03:17] s/it was versions/it was versioned/ [03:18] lifeless, 'bzr st' output http://pastebin.com/m532dfc0e [03:24] wildfire: i'm looking into that assertion [03:29] poolie, okay, I'll probably only be up for another 30 mins or so === jamesh_ is now known as jamesh [03:49] wildfire: i think you are hitting bug 150438 [03:49] Launchpad bug 150438 in bzr "AssertionError: Could not find target parent in wt in wt4 _process_entry" [Medium,Incomplete] https://launchpad.net/bugs/150438 [03:50] can you provide any moer information beyond the traceback about what caused it? [03:50] poolie, progsoc is using bzr to manage /etc on a few machines; I did a 'dpkg -P blosxom' which also removed the /etc/blosxom directory [03:51] poolie, I then wanted to commit that directory removal, so I did "bzr commit blosxom" and from this point onwards bzr is failing with an internal error on a number of commands [03:55] thumper: I would breakpoint in the failing tests setup(); and check that the queue setup's setUp() really is working correctly [03:55] lifeless: ok [03:55] lifeless: I'm just off to collect kids from school et al, and will get back to you [03:55] k [03:55] wildfire: thanks [03:56] to recover that directory i suggest you [03:56] bzr branch /etc /tmp/new-etc [03:56] mv /etc/.bzr /etc/.bzr.broken [03:56] mv /tmp/new-etc /etc/.bzr [03:56] (maybe make a backup of all of etc ot a tarball first) [03:57] poolie, actually I don't want to recover the directory. I want to record the removal of it ... [03:58] i meant, to get bzr back in a working state [03:58] ah, okay [03:58] to commit the removal of it and avoid this bug [03:58] i suggest you just remove it, and then commit the whole thing (with no other changes made) [03:58] I think the last line should be "mv /tmp/new-etc/.bzr /etc/.bzr" [03:58] thankyou for reporting it [03:58] ! [03:58] you're right, sorry [04:02] poolie, hmm, the first command 'bzr branch /etc /tmp/new-etc' is also failing with the internal errors [04:02] poolie, also, just in case this was related to the interative plugin I checked with that removed and it also fails [04:02] could you paste it? [04:02] sure, one sec. [04:02] spiv: ivar - thats ":ivar THING: details" right ? [04:03] lifeless: yeah, I think so. [04:03] just like param, except s/param/ivar/ :) [04:03] poolie, http://pastebin.com/m5ffc2800 [04:04] what does 'bzr log -r -1 /etc' show you? [04:05] wildfire@muspell:/etc$ sudo bzr log -r -1 /etc [04:05] ------------------------------------------------------------ [04:05] revno: 51 [04:05] committer: Anand Kumria [04:05] branch nick: Muspell /etc [04:05] timestamp: Mon 2008-04-14 11:56:12 +1000 [04:05] message: [04:05] Ignore generated berkeley db files [04:06] is that the revision you were trying to commit when it failed/ [04:07] does bzr branch -r -2 work? [04:09] bzr log -r -2 works [04:09] the revision I am trying to commit is the removal of the /etc/blosxom directory [04:11] ok [04:11] can i suggest you start from the branch of revision -2, put it back in place as above, and then try the commit from the top of your tree [04:12] poolie, you mean 'bzr branch -r -2 /etc /tmp/new-etc' and then go from there? [04:12] yes [04:15] poolie, done and now 'bzr status' tells me that the blosxom directory has been removed [04:15] ok [04:16] how should I now go ahead and commit that ('bzr commit blosxom' is what caused the breakage before) [04:16] i think you should just commit with no arguments from the root of /etc [04:16] poolie: I'm smelling an update_by_delta bug [04:16] unless you want to see if that reproduces it [04:16] ahh, I have other unrelated changes that I'd like to commit in a separate changeset [04:17] wildfire: bzr shelve++ [04:18] ah [04:18] it's possible those changes are causing this [04:18] could you paste bzr st? [04:18] one second [04:19] actually it would be easier if you just put it in bug 150438 [04:19] Launchpad bug 150438 in bzr "AssertionError: Could not find target parent in wt in wt4 _process_entry" [High,Confirmed] https://launchpad.net/bugs/150438 [04:19] since that's just what i was going to do :) [04:19] http://pastebin.com/d1e7a407 [04:20] poolie, I'll try but it appears progsoc's email is b0rked, which is why I'm trying to look at the configuration (and remove unnecessary stuff) [04:24] ok, thanks for that [04:24] lifeless: ^^ [04:24] it doesn't *look* strange [04:25] wildfire: if you want them in separate commits i suggest selectively committing everything but the blosxom deletion [04:27] poolie, just tried (bzr add .bzrignore; bzr commit .bzrignore; bzr status) which returns me to the broken state again [04:28] .bzr << bloxsom << dovecot [04:28] i need help commit with bzr [04:28] someone help me [04:29] lifeless, -EPARSE ? [04:29] wildfire: I'm speculating about the nature of the bug [04:29] hersonls: just ask your question please [04:29] poolie, how i send the commit to launchpad? [04:29] lifeless: you suspect another ordering problem in the dirstate? [04:30] oh, okay, well I can redo the tmp/new-etc thing and try the dovecot one first then [04:30] one sec [04:30] hersonls: have you created a Launchpad account? [04:30] yeah [04:30] poolie, i try bzr push and return locked 30 minutes, 24 seconds ago [04:30] hersonls: what url are you trying to push to? [04:30] poolie: in the update via delta code of commit; yes [04:30] poolie, bzr+ssh://hersonls@bazaar.launchpad.net/~hersonls/bzr/slackbuild [04:31] hersonls: did you maybe previously try to push to it and interrupt the process? [04:31] hersonls: if so ,try 'bzr break-lock URL' [04:32] lifeless, poolie: okay, redone. (bzr add dovecot/dovecot.conf; bzr commit dovecot/dovecot.conf; bzr status) works [04:35] wildfire: I speculate that add .bzrignore; commit; status will still break [04:36] lifeless, if you had bet, you would be [04:36] ... [04:36] wrong [04:37] :) [04:37] biab [04:38] fwiw you could have done 'bzr commit dovecot' - dovecot/dovecot.conf was already added according to your status [04:41] poolie, i push to the branch bzr+ssh://hersonls@bazaar.launchpad.net/~hersonls/bzr/slackbuild and in https://code.launchpad.net/~hersonls/ say be empty [04:43] hersonls: it doesn't look empty to me [04:43] lifeless: oh could you please fix 1.4 pqm this afternoon, if you did'nt already? [04:43] hersonls: there is a slight lag before the web ui updates, you might have hit that [04:44] poolie, i can lock this branch for push by another users? [04:45] hersonls: oh is this the data to build bzr in slackware? great! [04:45] poolie, yeah [04:45] hersonls: you want to let them write, or not let them write? [04:46] man, team creation is just impossible to find in lp... [04:46] not write [04:46] ok [04:46] i can do that? [04:46] because that branch is owned by you, only you can write [04:46] ohhh [04:46] yeah [04:47] i can build team for this branch? [04:47] on that page it says (to me) "Upload URL: You cannot upload to this branch. Only hersonls can upload to this branch. " [04:47] oh, tanks... [04:47] i brazilian, and my english is bad [04:47] sorry [04:47] if you want it to be writeable by a team create one at https://launchpad.net/people/+newteam [04:47] it's ok [04:48] de nada :) [04:48] ohh [04:48] você fala portugues? [04:48] un poco [04:48] uh that's about the limit :) [04:48] bomm [04:49] but there are several brazilian people in #launchpad [04:49] or indeed sometimes here [04:49] like kiko [04:49] biab [04:51] poolie, my team can write in branch? [04:53] you'd need to also change the branch to be owned by that team [04:55] ok [04:56] poolie, the list of all files of the branch, can see where? [04:56] in http://bazaar.launchpad.net/~hersonls/bzr/slackbuild [04:56] ? [04:57] hersonls: click the directory name [04:57] (really gone now [04:58] i can use that links to wikipage of the bazaar for download? [05:00] poolie, ? [05:29] bbiab [05:42] lifeless: found the problem [05:43] lifeless: ping me when you have some time to talk about it === mwhudson_ is now known as mwhudson [06:31] thumper: so [06:32] lifeless: bug 217112 [06:32] Launchpad bug 217112 in pqm "Test failures on a clean checkout" [Undecided,New] https://launchpad.net/bugs/217112 [06:33] lifeless: I was also bitten by 210165 [06:33] bug 210165 [06:33] Launchpad bug 210165 in launchpad-bazaar "BugBranch links created after pushing a copy of a branch with "bugs" revision properties" [High,Fix committed] https://launchpad.net/bugs/210165 [06:33] lifeless: so I'm cleaning the branch on LP right now [06:35] thumper: so, linking the branch is good [06:36] what would be better is linking the branch *and* providing a link which will show me the merge preview of the branch. [06:36] (presumably via loggerhead) [06:36] lifeless: the merge preview stuff is coming (RSN) [06:36] unfortunately I have to head out for an hour and a bit [06:36] so we could talk about this tomorrow morning if you like [06:37] cool; but it should be there *before* the merge is requested, if you see what I mean [06:37] thumper: +1 on the fix; but the comment isn't needed there [06:37] it is clear that it is the right thing to do [06:38] ok [06:38] * thumper away [06:38] drop the comment and update the branch, I'll merge it to mainline tomorrow [08:32] New bug: #217134 in bzr "Break-lock: Bzr doesn't accept my keystrokes to allow it to continue" [Undecided,New] https://launchpad.net/bugs/217134 [08:49] night all [08:54] thumper: well, a) lp should notify the bug subscribers [08:55] thumper: and b) a bug comment would be fine [08:55] night ian [08:55] yeah, I guess [10:34] Ran 783 tests in 29550.804s [10:56] New bug: #217180 in bzr-svn "Bzr update crashes" [Undecided,New] https://launchpad.net/bugs/217180 [11:47] New bug: #217188 in bzr "bzr update crashes with KnitCorrupt error" [Undecided,New] https://launchpad.net/bugs/217188 === mrevell is now known as mrevell-lunch [12:58] /win 12 [13:24] Hello! a question: I used bzr on a Windows machine, to modify a file originally created under Unix (lines terminated with LF), and after that the file had all its LF changed with CR-LF, and "bzr gcommit" showed all lines changed because of that. [13:24] I remember some other revision tool which handles that fine (it must discard CR-LF along the way); [13:24] ideally your editor should not screw them up [13:24] is there a switch in bzr to have that too? [13:25] but there is development in bzr to add that feature [13:25] bob2: I agree; but I'm not driving Visual Studio's roadmap :)) [13:25] but it does not afaik exist yet [13:25] bob2: this is good news! do you know a place where the planned development is tracked (a bug report, anything?) [13:26] http://bazaar-vcs.org/LineEndings seems still current tho old === mrevell-lunch is now known as mrevell [13:28] "Status of this work: I have working implementation and will send merge request soon.", yay [13:30] bob2: and that comment is 2 days old!! [13:30] bob2: thank you for the link! [13:30] no worries === pmezard_ is now known as pmezard [14:32] jelmer: is the 0.4 bzr-svn branch "safe" to use with bzr 1.4rc1 at the moment? or should I stick with bzr 1.3 for a while longer? [14:40] Anyone know the status on custom merge support? The particular case I'm interested in is "per filetype merge support" ; I want to be able to unpack an XML file into a more merge-friendly line format, then pack it back up again after the merge has finished. [14:50] * awilkins reads the code and concludes that he could make a plugin that implemented a merger that inherited from Merge3Merger or Diff3Merger that did the required stuff. [14:50] Would that be accurate? [14:53] awilkins: there is a plugin that implements diff for ms word files iirc [14:54] Aha, yes, I'll take a look at the source for that [14:54] http://gigo-ice.org/scm/bazaar/plugins/docdiff.html [14:55] * awilkins is quick on the draw and is aleady pulling the branch :-) [14:55] :) [14:57] Not quite what I wanted though ; I want something that meshes seamlessly with "bzr merge" for my special case. [14:58] I think I see the way though - if I inherit from one of the existing mergers I should be able to register a new one. [15:13] awilkins: That's correct. Odd_Bloke was working on support for custom merges at the sprint, but I haven't heard since. [15:14] abentley: From what I've read so far, the merge classes could be refactored out a little ot make custom merging a bit easier to implement [15:14] awilkins: I'm sure that's true. [15:15] abentley: e.g. ; all I require is a pre-process step and a post-process step, but I'm probably going to have to (at a minimum) cut/paste the whole text_merge routine from Merge3Merger [15:16] Odd_Bloke: awake? [15:32] howdy [15:33] i am having problems pushing a local branch using bzr 1.4rc1 to a server running 1.3.1 (i suspect this is since the upgrade, but am not certain). is that a known problem? === ja1 is now known as jam [15:46] intellectronica: no [15:46] abentley: cheers. i'm now discussing this with jam on bzrlp, b.t.w === `6og is now known as Kamping_Kaiser [16:01] New bug: #217290 in bzr "Error during "bzr commit" ERROR: exceptions.MemoryError" [Undecided,New] https://launchpad.net/bugs/217290 [16:44] abentley, awilkins: Just going to bed, in fact. To answer several questions at once, I'm not at uni ATM so am catching up with family and friends who I don't see that often. I head back on Wednesday and so should get back into the swing of things shortly. :) [16:44] * Odd_Bloke --> sleep the sleep of the dead === thekorn_ is now known as thekorn [17:00] New bug: #217313 in bzr ""bzr push" fails with error "unable to rename ..."" [Undecided,New] https://launchpad.net/bugs/217313 [17:49] I wonder what the reaction would be to a bzr plugin with an amusing name. I'm thinking of a combined push/pull, which syncs in either direction provided one side is just older than the other... "bzr sync" sounds too generic. What about "bzr llama", as a joke on the Push-me-Pull-you from Dr Doolittle === kiko is now known as kiko-fud [18:07] good morning.. what's the correct lp: equivalent for bzr branch http://bazaar.launchpad.net/~zope2book/zope2book/zope2book ? [18:09] I thought bzr help urlspec would tell me [18:09] * sm has 1.3 [18:15] why do I get bzr push lp:~zope2book/zope2book/zope2book [18:15] bzr: ERROR: Cannot lock LockDir(http://bazaar.launchpad.net/%7Ezope2book/zope2book/zope2book/.bzr/branch/lock): Transport operation not possible: http does not support mkdir() [18:15] ? [18:15] sm: you can't push to launchpad over HTTP [18:15] sm: use bzr+ssh instead [18:15] ah, thanks [18:16] sm: oh, I see your command. I think you can use 'bzr lp-login' and a sufficiently recent version of bzr to have that automatically happen for 'lp:' urls [18:16] or maybe I'm making all of that up [18:20] trying bzr lp-login username then push lp: .. it's sitting there for quite a while [18:20] * sm has made one small patch [18:20] worked! thanks [18:20] sm: cool! [18:20] I'm not a liar! yay [18:24] Hm. bzr viz misnumbering revisions... === kiko-fud is now known as kiko [19:30] New bug: #217377 in bzr "merge in shared repository failed assertion" [Undecided,New] https://launchpad.net/bugs/217377 [20:25] Hey guys, I'm still new to bzr - is there a way to cherrypick individual hunks out of a file for a commit? [20:25] I couldn't find anything on the wiki [20:26] so I figured I'l ask you guys. [20:26] oh, and I'm on 1.2.0 [20:31] dwt - bztools shelve plugin === mw is now known as mw|food [20:40] thanks [20:47] dwt: and the interactive plugin provides the opposite [20:47] awilkins: Thats indeed very cool. I do prefere to puth stuff to the side [20:47] so I can test in the tree [20:47] before comitting [20:48] However I don't quite get how to tease apart individual hunks. [20:48] Is that possible with shelve? [20:48] it is [20:48] it will show you each hunk [20:49] and ask what to do with it [20:49] No, you can't work at a sub-hunk granularity with it... [20:49] Is there a way to separate a hunk manually? [20:50] I mean, so bzr sees it as two hunks? [20:50] ah, I misunderstood the question [20:53] Hi all. Just wanted to say how much we love bzr. We just re-rooted our entire branch effortlessly and bzr handled all the merges like a dream. [20:56] dato: Thanks anyway! [21:18] dwt: At sub-hunk granularity, you're better off using your editor's diff mode. [21:18] e.g. bzr diff --using gvimdiff [21:19] abentley: Well, my diff tools all dont let me choose what parts of a hunk I want it to see as a new hunk. :-( [21:20] But maybe I'm missing something [21:21] dwt: For shelf to work at sub-hunk granularity, it would have to become a crappy editor. [21:21] Can I use bzr-svn on an svn repo over https? [21:21] So use an editor instead. [21:22] abentley: yeah, I could try to insert something that was in the old revision in between those so I can shelve the other stuff and then return to the previous state [21:22] dwt: What is your editor? [21:23] I use Xcode mostly [21:23] Or Textmate [21:23] On OS X [21:23] Does Textmate have a diff mode? [21:23] not really - it can view diffs [21:23] in very nice ways [21:24] though no commands to bzr from there [21:24] Ah. Vim and Emacs both provide ways to edit a file while showing a comparison to a previous version. [21:25] How does that allow me to split a hunk [21:25] ? [21:25] Which makes it easy to restore old lines. [21:25] ok, true [21:25] There is no hunk-splitting support in bzr because if you need to split hunks, you really need to use an editor anyhow. [21:26] This does not mean that the editor does hunk splitting. [21:27] It means that the editor is a better alternative than hunk splitting. [21:30] anybody any idea how long it usually takes for the OS X installer packages to be updated? === mw|food is now known as mw [21:35] abentley: Well, thanks for the help [21:36] I will see how that works. [21:56] dwt: there is a textmate plugin here, but I have not tried it myself: http://bazaar-vcs.org/TextMateBundle [21:56] statik: thanks for the link! === mwhudson_ is now known as mwhudson [22:30] New bug: #217454 in bzr-svn "Trying to branch a subversion branch into a shared repo gives AssertionError" [Undecided,New] https://launchpad.net/bugs/217454 [23:16] jelmer: does bzr-svn 0.4.9 work with 1.4rc1? [23:16] mxpxpod: Nope, there's nothing compatible with 1.4rc1 out yet [23:16] mxpxpod: I hope to release something before friday (when 1.4 is released) [23:16] jelmer: gotcha [23:32] jam: I am fixing the regression [23:32] lifeless: thanks [23:37] jam: did you file a bug? [23:38] lifeless: no, I didn't. Do you want an official bug? [23:38] no [23:38] just would have recorded if there was one [23:38] sure, how did you chose to fix it, btw [23:38] patience kemosabe :) [23:39] * Severe performance degradation in fetching from knit repositories to [23:39] packs due to parsing the entire revisions.kndx on every graph walk [23:39] iteration fixed by using the Repository.get_graph API. (Robert Collins) [23:42] lifeless: I had thought about that at one point, but it didn't make it into the email, and I also wasn't sure if Graph.get_parent_map() would also be cached. [23:42] But I guess KnitRepo.get_graph() returns a wrapper around the KnitVF itseldf [23:42] KnitRepository._make_parents_provider returns a closure with the revision versioned file [23:43] actually with my changes it can return the revision vf directly [23:43] lifeless: Implementing get_parent_map directly on KVF? [23:44] lifeless: have you been telling people about your fix on #bzrrlp? [23:44] already done [23:45] no, should I? [23:46] poolie: ping; I think versioned properties are likely a bad idea and not a dependency for the eol stuff bialix has done [23:46]