| exarkun | Uh, what's this supposed to mean? http://codepad.org/doDaxWiE | 13:38 |
|---|---|---|
| * exarkun goes to do dishes while waiting for the answer | 13:38 | |
| exarkun | aw, come on :/ | 13:57 |
| \u03b5 | Hmm I just had bzr try to write a change twice | 15:23 |
| \u03b5 | I pulled a revision from a work branch into a mainline branch bound to launchpad | 15:24 |
| \u03b5 | It gave an error that I had already acquired a lock on the repo seconds ago, but bzr update brought in the revision anyway | 15:24 |
| \u03b5 | I can't seem to reproduce that with only local branches | 15:25 |
| === maxb_ is now known as maxb | ||
| \u03b5 | hm | 15:28 |
| \u03b5 | now it applied cleanly on a test branch :( | 15:28 |
| * fullermd_ takes credit for fixing it. | 15:30 | |
| \u03b5 | that was quick, fullermd_! | 15:30 |
| fullermd_ | That's why I get paid the Big Bucks(tm). | 15:31 |
| \u03b5 | well damn it, I still get it on the real branch | 15:34 |
| fullermd_ | exarkun: If you use -Dhttp to trace the requests, you'll find the last one that works is grabbing .bzr/branch/location, which contains that filtered:/// string. Which I think means that the "branch" there is actually a branch reference (light checkout) or something else. | 15:34 |
| fullermd_ | exarkun: (no idea why it has a filtered:/// location to be sure, but at any rate, it's likely that whatever the path in there is, it won't make much sense over dumb http) | 15:35 |
| fullermd_ | \u03b5: Well, stop breaking it again! ;p | 15:36 |
| \u03b5 | I'm not sure what I can do to distinguish the two branches... | 15:37 |
| fullermd_ | What's the exact error it's throwing, and see if there's anything revealing in .bzr.log. | 15:38 |
| fullermd_ | (and I may not respond too quickly; I'm just running away from my computer for a bit right now) | 15:38 |
| \u03b5 | I can't find any *bzr.log* files in .bzr or ~/.bazaar, did it get deleted when I ran another command? | 15:40 |
| \u03b5 | nope, still no log | 15:41 |
| fullermd_ | No, in ~ | 15:41 |
| \u03b5 | what's the preferred pastebin here? | 15:41 |
| exarkun | fullermd_: Thanks | 15:43 |
| \u03b5 | http://paste.ubuntu.com/1265900/ | 15:44 |
| \u03b5 | scratch that, http://paste.ubuntu.com/1265901/ | 15:44 |
| exarkun | fullermd_: How do I use that -Dhttp thing? Neither "bzr -Dhttp pull" nor "bzr pull -Dhttp" shows me anything extra. | 15:47 |
| \u03b5 | the test branch commit looks similar, without the lock exception | 15:48 |
| \u03b5 | I'll try printing a stack trace on that log entry | 15:49 |
| \u03b5 | http://paste.ubuntu.com/1265936/ | 16:03 |
| exarkun | `JailBreak: An attempt to access a url outside the server jail was made:ยด? | 16:40 |
| exarkun | Server jail? | 16:40 |
| fullermd_ | \u03b5: Do you have write permission to the lp: branch? | 16:40 |
| fullermd_ | exarkun: I did "bzr -Dhttp info" on it. The trace drops in your ~/.bzr.log | 16:40 |
| \u03b5 | yes in fact the revision pushed fine | 16:40 |
| fullermd_ | Yes, the smart server jails itself. chroot()-like. | 16:41 |
| exarkun | fullermd_: Ah, I see, it did indeed log there, thanks. | 16:41 |
| \u03b5 | I didn't have it locally but bzr update brought it home | 16:41 |
| exarkun | fullermd_: I just changed my post change branch tip hook from doing a lightweight checkout to doing a regular checkout. The lightweight checkout worked fine, but the regular one is a jail break attempt? | 16:42 |
| fullermd_ | exarkun: Got me. I'd guess at some oddity in the env happening around the hook (the heavy would try to potentially write into the repo frex), but that's well outside of my ken. You'd need somebody who knew internals better for that. | 16:43 |
| exarkun | Okay, thanks. | 16:43 |
| exarkun | Where _is_ writing allowed? | 16:43 |
| fullermd_ | \u03b5: Mmm. Well, in the earlier trace it looks like you're pull'ing into a checkout (heavy probably) of the lp: branch, and something in that process is getting screwy when it tries to write the revs up to LP. | 16:44 |
| * exarkun bets it's allowed at a location that's undiscoverable | 16:44 | |
| fullermd_ | Wherever it got jailed at. In the case of bzr://, it's whatever dir you aimed bzr server at. For bzr+ssh, I think it's at the FS root. | 16:45 |
| exarkun | I'm using bzr+ssh, and I tried to do a checkout to a directory in ~, which is a sibling of the branch location | 16:45 |
| fullermd_ | But again, this is really pushing the bounds of my knowledge. It may be a somewhat spurious result too, of just doing something unexpected deep inside the stack. | 16:45 |
| * exarkun nods | 16:45 | |
| fullermd_ | You're pretty deep in the middle of commit/push/whatever with that hook. I'm not sure anybody's tried doing a lot of fiddling down there before, so... | 16:46 |
| \u03b5 | My outsider guess would be that pull is trying to push the branch while the local bound branch already knows to do that | 16:46 |
| exarkun | :( | 16:46 |
| quotemstr | How do I get bzr to give me the original hunk in addition to TREE and MERGE-SOURCE? | 20:27 |
| === thomi is now known as thomj | ||
| === bradm_ is now known as bradm | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!