=== yofel_ is now known as yofel [03:17] jelmer: why did you marjk lp:~jelmer/testrepository/fix-uxsuccess rejected? [04:50] hmmm [04:50] anyone know what this error means? :< [04:50] bzr: ERROR: Failed to rename C:/Users/Mark/Code/bzr/mips/csc252/trunk/prog2 to C:/Users/Mark/Code/bzr/mips/csc252/trunk/.bzr/checkout/limbo/new-5/prog2: [Error 5] Access is denied === nyuszika7h is now known as nyan_cat === nyan_cat is now known as nyan-cat === nyan-cat is now known as nyuszika7h [11:55] lifeless: I got a disapprove vote and I'm not going to look into it further at this point. === Quintasan_ is now known as Quintasan [14:53] hi jelmer [14:53] hi Noldorin_ [14:53] how's it going? [14:54] alright, how are you? [18:59] jelmer: I thought it was just needs-tests [18:59] jelmer: jml said he was going to look at landing it himself [19:00] lifeless: yes, but with tests - my fix was trivial [19:05] lifeless: anyway, I was mainly just trying to clean up my list of in-progress branches [19:06] lifeless: I agree it wasn't really rejected as such, but there isn't really a way to stop being involved in a MP otherwise [19:07] rejection doesn't stop involvement either :) [19:08] lifeless: ok, you got me there. but it at least means it no longer shows up in my wip and needsreview listings [19:08] sure. I can understad the motivation === marienz_ is now known as marienz === marienz_ is now known as marienz === marienz_ is now known as marienz [22:36] bzr: ERROR: exceptions.AssertionError: get_next() called when there are no chars left [22:36] sigh [22:36] that's in 2.2.4-0ubuntu1 [22:44] lamont: hi [22:44] lamont: that's usually a sign of a corrupt dirstate because of a truncated file - unsafe unmount? [22:45] lamont: FWIW, newer versions of bzr fdatasync more agressively [22:46] jelmer: machine panic [22:46] so yeah, that could be consistent with an unsafe umount [22:46] OTOH, it should be ext3 [22:47] lamont: removing the checkout and creating it again should fix, but will discard any pending metadata changes you had in bzr [22:47] "rm -rf .bzr/checkout; bzr co" [22:48] something tells me that a bzr co might tend to modify files in the working directory [22:49] lamont: IIRC it's not impossible for the "write tmp file, rename tmp to final location" idiom to leave 0-byte files after a crash with ext3, because by default it allows metadata to be committed before data. [22:50] spiv: and whether or not that is a grave bug in ext3 has been debated before [22:50] The O_PONIES feature request :) [22:53] hey spiv :) [22:53] o/ [22:56] spiv: thats ext4 I believe [22:56] spiv: ext3 cna't selectively sync journal contents [22:57] lifeless: IIRC it's possible (but less probable) under ext3 too [23:00] spiv: you need journal_data_writeback for that [23:00] spiv: IIRC which is not on ext3 by default (it defaults (or defaulted?) to journal_data_order [23:01] spiv: IMBW - I haven't checked the code since ages ;) [23:02] lifeless: http://lwn.net/Articles/351422/ says the defautl mode under ext3 switched from data=ordered to data=writeback around 2.6.30 [23:03] Possibly that change got reverted later due to this sort of fallout? [23:03] ah! heh [23:04] But anyway that's enough for me to stop short of being certain that "default ext3" == "necessarily safe for that idiom" [23:04] so if they changed that, then yes, its unsafe [23:04] and sigh, still no user space barriers [23:04] to we have to choose between slow and crap [23:05] or use btrfs! [23:07] hi lamont, spiv, jelmer, lifeless [23:11] howdy [23:11] there is a repair-dirstate operation [23:12] 'bzr repair-workingtree' [23:14] morning poolie [23:15] /1/1 [23:15] >.> [23:18] lamont, i think that's https://bugs.launchpad.net/bzr/+bug/709158 btw [23:18] Ubuntu bug 709158 in Bazaar "bzrlib._dirstate_helpers_pyx.Reader.get_next: AssertionError: get_next() called when there are no chars left" [High,Confirmed] [23:18] thanks for pointing it out [23:19] ah, cool [23:30] g'morning poolie [23:38] Morning. [23:39] wgrant@lucid-test-lp:~/launchpad/lp-branches/separate-zopeless-mail$ bzr pull :push [23:39] bzr: ERROR: Unsupported protocol for url "lp:~wgrant/launchpad/separate-zopeless-mail" [23:39] Is that meant to not work? [23:39] I believe the push URL is set in locations.conf. [23:40] wgrant: it's a known issue; bzr itself always sets expanded URLs (not requiring directory services) in configuration files [23:40] jelmer: Hmm, but bzr push works fine. [23:41] wgrant: does "bzr push :push" work? [23:41] Indeed not. Good point. [23:42] At least there is a workaround, then. Thanks! [23:45] wgrant: anytime === yofel_ is now known as yofel