[00:01] lifeless: fwiw, it's not a dpush bug [00:02] lifeless: it's a bug in the svn mac installer, which I've reported and yelled about in here like 5 times [00:02] https://bugs.launchpad.net/bzr-mac-installers/+bug/621446 [00:02] please, somebody fix this [00:02] Launchpad bug 621446 in Bazaar Mac Installers "AttributeError: paths when trying to svn-import (affected: 6, heat: 46)" [Undecided,New] [00:02] it's shipping with a prerelease bzr-svn that's broken. [00:02] ahh [00:02] exarkun: ^ [00:03] glyph: perhaps you could drop a mail to the list; its more likely to catch the mac installer dudes eyes [00:03] * exarkun arrives [00:03] btw the launchpad login experience sucks [00:04] exarkun: what are you talking about, it's great. It is super secure. What if you accidentally informed launchpad.net of your launchpad.net credentials? that would be terrible! [00:04] (sorry, apparently in a snarky mood tonight) [00:06] also, I may have asked this one before [00:06] but is there any way to change qbzr's fonts? [00:06] the code font it has selected for me is painfully ugly [00:07] glyph: no idea about qbzr, sorr. [00:07] exarkun: what about the experience sucks? [00:08] lifeless: I suspect exarkun is complaining about the same part of the experience that I'm snarking about. [00:08] lifeless: The "did you actually mean to log in" page that comes after the log in page [00:09] I'm on launchpad.net. I click 'log in'. I type my username and password, and hit 'continue'. It then says "The following information will be available to Launchpad:", but clearly that information is _already_ available to Launchpad. [00:09] Also, at the top of that page it says "Logged in as ..." even though I am (apparently?) not actually logged in [00:09] Personally, having worked on federated authentication, I understand the difference between authenticating to the login service and relaying that authenticating to the application, but that is an extremely subtle distinction [00:09] that I think is lost even on most programmers. [00:16] mmm [00:16] so it might be useful to file a bug or two here [00:16] we are working on allowing any provider [00:16] - becoming a general rp [00:19] lifeless: That would be great. [00:20] lifeless: I really, really want LP to be a general RP so I can throw these credentials in the garbage, along with all of my other random web credentials [00:20] yeah [00:20] lifeless: actually, is there a launchpad bug for that? I'd love to be notified when it's ready. [00:20] sure is [00:21] but still, in that case, I want the LP login service to special-case the case where I am logging in *to* launchpad *on* launchpad :). [00:21] well [00:21] LP doesn't have its own provider anymore [00:21] login.launchpad.net looks like a provider [00:21] it does [00:21] its login.ubuntu.com themed [00:21] different DB [00:22] lifeless: it says "launchpad" at the top, that makes me think it's launchpad ;) [00:22] I know [00:22] we had one for a bit [00:22] then folk used it [00:22] then we move it to a different department ;) [00:22] le sigh. [00:23] lifeless: when I said "is there a launchpad bug", what I really meant was" what is the URL for that launchpad bug" ;-) [00:23] I don't know [01:10] lifeless: well, I spent about 20 minutes searching, and I found nothing. So: [01:10] Launchpad bug 672004 in Launchpad Foundations "launchpad should be an openid relying party (affected: 1, heat: 6)" [Undecided,New] [01:21] glyph: https://bugs.launchpad.net/launchpad-foundations/+bug/210943 [01:21] Launchpad bug 210943 in Launchpad Foundations "be an openid consumer (affected: 32, heat: 221)" [Low,Triaged] [01:21] glyph: its on the first page of a search for openid in launchpad [01:21] glyph: I'm sorry you spent so long looking [01:23] lifeless: not for me it isn't [01:23] "No results for search openid" [01:24] http://bit.ly/9LDsKk [01:24] https://bugs.launchpad.net/launchpad-project/+bugs?field.searchtext=openid [01:24] oh, I already found this bug [01:26] ah [01:26] not launchpad itself [01:26] or launchpad foundations [01:26] but launchpad project :) [01:26] we use launchpad badly [01:26] sorry :( [01:26] (ok sorry for talking about launchpad so much in #bzr, I will go back to talking about bzr in here and join #launchpad if I want to talk about launchpad) [01:26] :) [01:32] so, regarding qbzr [01:32] there are a couple of places where it hard codes "Courier New, Courier" [01:32] and you can just edit it [01:34] and https://bugs.launchpad.net/qbzr/+bug/619301 is a bug about fonts, albeit maybe there should be more bugs [01:34] Launchpad bug 619301 in QBzr "Commit dialog: Font should be choosable (affected: 3, heat: 15)" [Undecided,New] [01:41] Can I reconstruct a Subversion repository from a bzr-svn branch of its root? ☺ [01:49] luke-jr_: No, as the zeroth revision in svn is immutable. [01:58] jelmer: Can't you rebase onto an empty trunk? [01:58] oh, wait, maybe not. [01:59] luke-jr_: Why bother, though? If you are in the happy situation of having lost your SVN repository but still having a bzr backup, sounds like it's time to just switch to bzr :) [01:59] jelmer: svnadmin can load over it ☹ [01:59] glyph: bzr doesn't work nicely with subtrees [02:00] glyph: I'd want to at least import it back to Subversion, then branch from the trunk [02:00] luke-jr_: I think you might be able to do a push if you use svnadmin to change the uuid to match the original repo first, and get rid of your bzr-svn cache for that repo [02:00] hmm [02:00] any idea how to find the UUID? ☺ [02:00] it's part of the bzr revision id [02:03] hmm, does seem to be working [02:03] any idea if this is lossless? :P [02:03] or at least keep compatibility with the real original repo, in case it returns someday? [02:03] luke-jr_: yes, this would be lossless in terms of what bzr supports (otherwise bzr-svn would refuse to push) [02:04] so svn:externals and other custom properties would be lost but svn:executable will be kept [02:04] i c [02:04] jelmer: I mentioned this yesterday but just got around to filing a bug on it: [02:04] so only lossless from Bazaar's limited view [02:04] Launchpad bug 672016 in Bazaar Subversion Plugin "'bzr branches' is pathologically slow with an svn+ssh URL (affected: 1, heat: 6)" [Undecided,New] [02:05] but my main purpose is that I am trying to write a little shell script that synchronizes a whole repository, including all of its branches. [02:05] I'd like it to work with a bzr repo or an svn repo. [02:05] 'svn ls' isn't quite right if you've got things which aren't branches in /branches, so it would be nice if bzr branches worked right. [02:05] is there a different suggested method for doing this? [02:08] glyph: hmm [02:09] * jelmer tries locally [02:16] glyph: it looks like this is an issue with the way the branches command works [02:17] it just find all possible places with a control directory and then filters out the ones that have branches [02:17] in svn, every directory is a possible control directory [02:20] jelmer: While it's theoretically possible to have a branch within a branch, it seems like you'd be pretty hosed if you tried that [02:20] glyph: it's perfectly possible in bzr [02:20] really? [02:20] well, the branch inside the other branch wouldn't be versioned as such, but it can exist. [02:21] (ie the container branch wouldn't know about it) [02:21] right [02:21] Well, certainly this is up to you, but personally I think it would be fair to say that anything which has the metadata to suggest that it's a branch is the end of the line for searching for more branches :) [02:22] and you could possibly have some configuration to disagree with that, for pathological repositories [02:22] glyph: it's not up to me :-) [02:22] aren't you mr. bzr-svn? [02:23] glyph: I guess :-) But bzr-svn doesn't get into the picture here except to answer the question "does this control directory have a branch attached to it?" [02:23] so there's no way I can make it scale from within bzr-svn, bzr core has to be changed. [02:24] jelmer: Doesn't bzr-svn also get asked "is this a control directory?" [02:25] glyph: Yes, but control directories can exist without branches. [02:26] jelmer: It should be possible to at least look at the directory and say 'no' if it was copied from somewhere else (i.e. probably a branch) [02:26] at any rate this is implementation, all I want is for 'bzr branches' to be super fast, just make it happen, etc ;-) [02:30] glyph: I don't think copies are a good indication of whether something is probably or not probably a branch. Also, finding out will have a performance impact too. [02:30] I'd like to fix this properly, by having bzr ask bzr-svn for the list of branches in a repository. I'm just not sure yet what the proper API for that is. [02:33] jelmer: My understanding is that SVN's revision calculus makes it formally impossible to determine the list of branches in the repository. [02:33] All you can do is look to see if a directory was copied from 'trunk' or something that looks like it. Is that wrong? [02:34] glyph: yes, but bzr-svn has a mechanism for determining what is a branch and what isn't. [02:34] glyph: that's what e.g. "bzr svn-import" uses to determine what branches to import. [02:36] jelmer: oh. I always assumed that _was_ the mechanism. What's the actual rule? [02:37] glyph: it uses heuristics to find out what repository layout a svn repository has and then uses that layout, which is usually path-based. [07:02] mgz: http://washort.twistedmatrix.com/2010/11/unicode-in-python-and-how-to-prevent-it.html === oubiwann is now known as oubiwann-away === oubiwann-away is now known as oubiwann [18:03] ...how do I tell which revision my branch is currently at? Neither 'status', 'log' or 'info' tells [18:03] bzr revno [18:03] or bzr log -r -1 [18:03] thanks :) [18:24] loldrup: if you've done a bzr update -r then you'll need bzr version-info to get it [18:31] gthorslund: bzr revno --tree [18:37] lifeless: aha! I've missed that one. thx === luks_ is now known as luks [20:48] so, I'm getting on a plane in 15 minutes [20:48] \o/ [20:49] can anyone figure out why svn+ssh://svn.twistedmatrix.com/svn/Twisted/branches/modules-so-2871 thinks that it needs to branch 14341 revisions, in a shared repository with an up to date copy of svn+ssh://svn.twistedmatrix.com/svn/Twisted/trunk [20:49] in less than that amount of time? [20:50] (svn+http should work as an anonymous mirror of the same thing) [20:50] well [20:50] jelmer: probably can, but I'd expect tha tmost of the history is identical and the progress counter should do a huge jump and be done quickly. [20:51] well, no [20:51] I know what it looks like when it does that [20:51] ok [20:51] and it's definitely not doing that now [20:51] don't stop the operation [20:51] 4519kB 15kB/s | copying revision 1126/14341 [20:51] oh [20:51] oh I'm a liar [20:51] but what does 'bzr missing svn+ssh://svn.twistedmatrix.com/svn/Twisted/branches/modules-so-2871 say, run from your trunk mirror [20:52] I have an up to date copy of http://svn.twistedmatrix.com/bzr/Twisted/trunk [20:52] who knows what the differences are there [20:52] if its the same bzr-svn version, it should be the same-same. [20:52] OTOH perhaps its building its bzr-svn cache first, if this is the first direct to svn operation you've done. [20:52] that can take a little time. [20:54] jam: oops, i wish i hadn't subscribed to ~jameinel/launchpad/lp-service :-) [20:55] lifeless: nope, not that either. I just made sure that I did have an up-to-date trunk; branching trunk took no time at all [20:55] there's something wonky with that branch [20:55] I'll have to repro this when I have better network connectivity and more time, I guess. [20:55] Maybe I'll get lucky and the plane will have wifi ;) [20:56] good luck. [20:57] lifeless: thanks for the attempt at help, anyway :) [20:57] :P [20:58] glyph: if the branch was branched at the wrong level in svn, it can behave like that [20:58] mwhudson: it wasn't; it's right under /branches, it's a copy of trunk [20:58] mwhudson: I mean, you have the appropriate credentials, go ahead and inspect it :) [20:58] that's unlikely if combinator is being used [21:01] glyph: well, i'm trying to reproduce at leat [21:01] *least [21:02] glyph: ok yeah, i see the same [21:06] oh hang on, different versions of the svn mapping in effect? [21:09] glyph: it seems lp:twisted contains a mix of the v3 and v4 bzr <-> svn revid mappings [21:12] and when you import from svn, you just get the v4 mappings [21:18] mwhudson: a mix? That seems odd. [21:19] Aren't they unrelated ancestries from bzr's point of view? [21:20] spiv: it does a bit doesn't it? [21:26] and it seems that bzr-svn is somehow smart enough to know that [21:26] bzr log --show-ids svn+ssh://svn.twistedmatrix.com/svn/Twisted/trunk -r 14639 [21:26] should give a v3 revid [21:26] but not smart enough to know that svn+ssh://svn.twistedmatrix.com/svn/Twisted/branches/modules-so-2871 -r 14639 should too [21:27] ... no [21:28] oh right [21:28] yes [21:29] i guess this is a bzr-svn bug === nailuj24 is now known as yailuj24 [21:37] mwhudson: how does bzr-svn know that 14639 should be svn-v3? [21:37] spiv: magic jelmer sauce? [21:38] spiv: actually, i guess that because there are revisions in tm.com svn that are actual bzr commits, this somehow forces all older commits to be of a known mapping [21:46] hmm! [21:49] * mwhudson wants to know what bzr log --show-ids svn+ssh://svn.twistedmatrix.com/svn/Twisted/branches/modules-so-2871 -r 14775 says, but it's taking freaking ages [21:49] ah [21:49] bzr: ERROR: Requested revision: '14775' does not exist in branch: svn+ssh://svn.twistedmatrix.com/svn/Twisted/branches/modules-so-2871 [21:50] i think maybe what's happening is that the revision that 'forces' the trunk into v3 mapping isn't present in modules-so-2871 [21:59] mwhudson: ah, that seems plausible [22:00] i don't know what you could do about that [22:02] nothing pleasant, I suppose... [22:03] Well, Twisted could land branches faster? ;) [22:03] yeah, that would definitely help [22:26] mornin all :) [23:07] thanks for that link spiv. [23:39] mgz: you're welcome! I haven't actually tried ascii_with_complaints yet myself... I suspect the noise will be a bit overwhelming. [23:43] I think part of the pain is there's no decent way to specify interfaces. [23:44] As in “the first arg of this function takes unicode not bytes”? [23:44] every time you write a function or method that takes some kind of string, you don't want to write a type check at the top. [23:45] as generally that should have already been handled somewhere else anyway. [23:45] Yeah. [23:45] Well, it was handled somewhere else... [23:45] That somewhere else being ~1970, where only ASCII mattered ;) [23:49] hi folks, i'm a bit confused, i have a branch on my local machine of a repo on my server, i also hhave a branch on the server for deployment, but i also made a sql backup to the server branch and checked that in on the server, but i can't get the updated sql back down on my local machin.e [23:51] hmm, never mind, backupappeared after i pushed tothe server...??? werid