/srv/irclogs.ubuntu.com/2012/01/25/#bzr.txt

=== lamont` is now known as lamont
Noldorin_hey folks00:49
Noldorin_is poolie back yet? :-)00:49
pooliei am here00:50
Noldorin_poolie, hi00:54
Noldorin_i'll try and catch you tomorrow, as it's a bit late here00:54
Noldorin_was just wondering if you were back from holidays00:54
Noldorin_had some dev related questions00:55
Noldorin_regarding some time ago00:55
Noldorin_ta00:55
vilahi all !08:28
vilapoolie: around ?08:28
pooliehey08:32
vilahurrah !08:32
pooliejust had a chat to thomas, all is well08:32
vilapoolie: looking at your udd patch for list_packages08:32
pooliethanks08:33
vilapoolie: or rather, since I approved it, looking at deploying (shutting down the importer right now)08:33
pooliei also would have guessed that it was a change on lp08:33
pooliethus my original bug title08:33
pooliehard to know08:33
vilapoolie: long story short: there are recent import failures for pristinetar, did you requeue these manually ?08:34
vilapoolie: if not, end of the interruption, if yes, I won't have to search why they were requeued ;)08:35
vilapoolie: ?08:49
poolievila, i did not do that08:55
vilak, thanks08:55
poolie...08:55
pooliemaybe at the rally?08:55
poolieit's a blur08:55
pooliei don't think so08:55
vilanope, more recent08:55
vilano big deal either08:55
poolielet's check shell history :)08:57
pooliethough, really, it ought to log something08:57
vilayeah...08:58
vilaI'm rsync the logs and will look at the package ones08:58
vilarsync'ing08:59
vilajelmer: quilt not installed breaks the importer (I just deployed bzr-builddeb 2.8.2) :-/ Will file a rt ticket, what is needed ? Just quilt or some more ?09:06
poolievila, as well as filing an rt, nag the vanguard in #is09:07
vilapoolie: will do09:07
vilabut usually when I do that I'm pushed back to file an rt ticket ;)09:08
poolievila, or, finish converting it to juju :/09:08
viladoing things in order will hopefully helps :)09:08
poolie;)09:08
vilapoolie: list_packages succeeded \o/ (no new packages apparently)09:11
pooliegreat09:12
vilajelmer: does your commit hook for quilt needs to crash when quilt is not installed ? Can't it disaplay a far red blinking warning instead ?09:13
vilajelmer: additionally, is there a config option to disable it (or any other way) ?09:13
vilajelmer: and finally, is there a minimum version required for quilt ?09:14
pooliegood night all, that's it for me09:30
vilagood night poolie !09:34
vilajelmer: rt filed, quilt installed, imports succeeded, want some names to check it worked as intended ?09:34
mgzmorning09:45
vilaheya mgz !09:50
vilajelmer: pakages that failed without quilt and have now succeeded: http://paste.ubuntu.com/816262/09:51
vilajelmer: quilt-0.48-5 has been deployed on jubany09:52
vilajelmer: and look at that... I was wondering how to avoid nexuiz-data endlessly re-trying while we investigate while it seem to be in an endless loop and it too failed with quilt, problem solved ;)09:53
rcsheetsI am getting an operation not permitted error when trying to push. http://pastebin.com/m7JvUbRv10:10
rcsheetsthe file referenced in the error message does not exist.10:10
rcsheetsin the branch i'm pushing *from*, that file (and the entire directory it's in) has been moved up a directory. perhaps that's confusing bzr?10:12
mgzrcsheets: hm, it really doesn't exist? I'd expect ENOENT rather than EPERM then10:18
mgzthere were some changes to chmod on trunk recently to ignore errors, but that was related to running against ntfs and other filesystems that don't support unix file modes10:19
rcsheetsthis is on ext410:20
rcsheetsstat: cannot stat `/srv/bzr/ffm-gms/trunk/cron/cron_customer_order_daily_email.php': No such file or directory10:20
rcsheetsactually i just noticed that from the destination branch, looking at 'bzr log', it seems like it worked10:21
rcsheetsbut i can't do an update10:21
vilarcsheets: which directory doesn't exist exactly ?10:21
rcsheetsPart of the most recent set of changes was moving public/cron/ to cron/10:22
vilaand which bzr version are you using ?10:22
rcsheetsbzr 2.2.410:22
vilahmpf, good to know, not tempted to upgrade to our latest stable that is 2.4 ? :-P10:23
rcsheetsi'm stuck for the moment on ubuntu maverick, but i can install a backport or just throw a new version into /opt10:23
vilak10:23
mgzokay, so it's being created as part of the merge, then can't be chmoded10:23
mgzI'd check the perms on the parent directory of the branch you're pushing to10:24
vilaI would start by checking the permissions of all directories10:24
vilamgz: :)10:24
rcsheetsi will chown them all to myself10:24
vilahold on10:24
* rcsheets holds on10:24
vilacan you check first so we have a reasonable understanding ?10:24
rcsheetsyes10:24
rcsheetsthe /srv/bzr/ffm-gms/trunk directory itself is owned by another user, but mode 775 and i'm in the group.10:25
vilaand below that ?10:26
rcsheetslooking10:26
vilahey bialix !10:26
bialixhi vila!!!10:26
bialixhi mgz!10:26
* mgz bops bialix 10:26
* vila looks at his zoo postcard ;)10:26
bialix:-D10:26
rcsheetsseveral other parts of the tree are owned by another user, but again i'm in the group and there are group write permissions10:27
* bialix is happy to say hello to bzristas10:27
vilabialix: :)10:27
rcsheetsthere is nothing in the whole tree below /srv/bzr/ffm-gms/ that i cannot write to10:28
mgzrcsheets: the other possibility is... the cron directory is getting created on merge? because you moved it?10:28
vilarcsheets: weird, try chown'ing and retry but I'm not fully confident it will work (on the other hand, I'm known for being rather pessimistic for bugs I don't fully understand ;)10:28
rcsheetsok10:28
rcsheetsi will tar up what i have now, in case i need to restore it later10:28
vilamgz: and ? ( a move requires write access on both dirs but that's still write access)10:29
vilarcsheets: did you check perms under the '.bzr' directory ?10:29
rcsheetsvila: yes i checked for anything not owned by me, using find. it found many things under .bzr owned by the other user, but with group write access.10:30
rcsheetsgroup=admin throughout the directory structure, and i am in that group10:30
vilagreat10:30
rcsheetsnow that i own everything, it seems to work fine10:31
rcsheetsyeah, that did it10:31
vilaweird10:31
rcsheetsindeed. well, thanks!10:31
vilauntil now, the issues we had with group perms always involved an sftp server ignoring bzr requests for group bits10:31
vilaunless fullermd remembers otherwise... never overestimate alzheimer ;-D10:32
rcsheetsi can retry using the latest bzr from my backup of the repository, if it would be helpful10:32
mgzbialix: there are a few qbzr/bzr-explorer things I think that need to happen before 2.5, I wonder if you have any other suggestions10:34
mgz1) make filesystem watcher disabling work, and make it only watch versioned files10:34
vilarcsheets: it's an issue with the working tree so it will useful only if you backed it up with the right perms10:35
mgz2) fix some exception encoding stuff once bzr has a mechanism for it10:35
mgz3) check if colo branches break anything too badly10:35
rcsheetsvila: i backed it up using "tar cf file.tar directory" as root. i *think* tar defaults to preserving all the ownership/permissions if you're root, but i may be wrong.10:36
mgzI'll do some of that at least next week.10:36
vilarcsheets: nope, that's my belief too10:36
rcsheetsit looks like it preserved everything10:37
rcsheetsi'll untar it somewhere else10:37
fullermdThe thing with sftp was that it filtered out the setgid bit.10:37
vilarcsheets: file a bug then so we can keep track of the issue10:37
rcsheetsvila: i thought i'd check first to see if the bug still exists in the most recent release.10:38
vilafullermd: yup, but did you ever encounter a case where the group bits were not respected while *not* using an sftp server ?10:38
vilarcsheets: yup, exactly10:38
rcsheetsk10:38
fullermdNot I.  But I wouldn't hit an issue even then, since I use a sane system that doesn't need that setgid crap anyway  ;p10:38
vilahehe10:38
vilamgz, bialix: from the top of your joined heads, how does the bzr installer build the ccurl-ca-buble.crt file and where is it stored ?10:40
vilageee10:40
=== bialix_ is now known as bialix
vilacurl-ca-bundle.crt10:40
vilabubble...   lovely tyop10:41
bialixvila: I've got it from mozilla10:41
vilabialix: as in: got in once and never rebuild it ?10:41
bialixit was stored along bzr.exe or inside lib/10:41
bialixvila: well, I updated it several times in 2007-2008 I think when I maintained windows build10:42
vilabialix: ok10:42
bialixmgz: I was disconnected, if you said something before vila asked us about cert, please, repear10:43
bialixmgz: I was disconnected, if you said something before vila asked us about cert, please, repeat10:43
rcsheetsvila: everything worked fine using 2.5b510:43
vilarcsheets: \o/10:44
bialixvila: what's up with cert?10:44
vilabialix: the urllib implementation can now verifies certs as long as some root certificates are provided... same issue you had with pycurl, I'm trying to gather how to find them for all platforms now10:45
RiddellI just noticed that debian bug 641019 is fixed10:45
ubot5Debian bug 641019 in pristine-tar "pristine-tar does not work with tar files made by openSUSE" [Normal,Fixed] http://bugs.debian.org/64101910:45
vilaRiddell: hello !10:45
rcsheetsfixed bugs are the best kind of bugs10:45
Riddellso UDD should be workable with KDE tars, I guess it needs a sync10:45
vilaRiddell: yup, I've filed rt #5045510:45
bialixvila: right. because bzr switched to urllib as default and it did not check certificates, I think that's why cert bundle wasn't updated for a looong time10:46
vilabialix: ???10:46
Riddellvila: lovely10:46
vilabialix: urllib will be the default httpS implementation in 2.6 only10:47
bialixvila: long time ago in far far galaxy10:47
vilabialix: are you saying the windows installer don't carry pycurl anymore ?10:47
viladoesn't10:47
Riddellvila: I'll look at getting p-t 1.16 into precise when I get a chance10:47
bialixI think so10:47
bialixvila: yep, exactly10:48
vilaRiddell: that's lucid we need for jubany but getting >= 1.17 in precise won't hurt !!10:48
* vila coughs10:48
vilamgz: ^^10:48
Riddellyeah, needs backporting too10:49
vilaRiddell: I didn't check if 1.16 was enough but 1.17 seem to carry additional fixes10:49
mgzsorry, was off responding to naoki's mp, unfortunately that's just the kind of change that won't work10:49
* mgz catches up10:49
mgzbialix: will paste what I said in /query10:50
bialixmgz, Riddell: I fear new filesystem watcher in explorer do more harm than help10:51
mgzheh, I bet the certs we ship still have that dutch registrar in then.10:52
bialixmgz: I think the best we can do before bzr 2.5 is out is make it disabled by default and provide config option for brave hearts10:52
mgzthere are six or so bug reports that I think are filesystem watcher related10:53
bialixmgz: yes :-(10:53
Riddellbialix: more problems from it?10:53
mgzit's mostly down to two things,10:53
bialixRiddell: mgz>there are six or so bug reports that I think are filesystem watcher related10:53
mgzit watches *everything* in the tree, including the .bzr dir, which can confuse other parts of bzr10:54
bialixon windows it triggers Access is denied error tooooo often10:54
Riddellbialix: I found it a big help for using explorer, if I have to use refresh all the time I find it no better than command line10:54
mgzand on windows watching counts as a soft-lock10:54
bialixthat's limitation of filewatcher in Qt10:54
Riddellbialix: but if there are problems then maybe it's more bother than it's worth10:54
mgzI think that just limiting to versioned files should fix nix at least, then we can leave enabled there10:55
bialixRiddell: I think we hit the limitation barrier of Qt itself here10:55
Riddellto be fair on Qt it is limited by the operating system :)10:55
mgzthere are ways around the windows issues, eg10:55
mgzmoving a file in explorer should unwatch any children first, then rewatch after moving10:56
mgzbut just making the disable/enable work should be the first thing10:56
bialixRiddell: right10:57
RiddellI'd love to help but I'm behind on kubuntu stuff because this accident I had has made me slower so I doubt I'll get to it10:58
mgzRiddell: your occasional opinion is all I'll bug you for :)10:58
bialixme too10:59
mgzis there anything else we think has to be done for qbzr or bzr-explorer before 2.5?10:59
bialixmgz: back to your list: what do you mean by 2) fix some exception encoding stuff once bzr has a mechanism for it10:59
bialixmgz: and 3) check if colo branches break anything too badly -- do you mean new colocated support in the core from jelmer?11:00
mgzah, er bug 599860 and any other things that will need fixing to once bzrlib changes how it does some trace encoding stuff11:00
bialixbecause old bzr-colo stuff is used to work11:00
ubot5Launchpad bug 599860 in Bazaar Explorer "UnicodeDecodeError with BzrError containing localised error message from OS" [Medium,Confirmed] https://launchpad.net/bugs/59986011:00
mgz^and right, jelmer's core colo things11:01
vilamgz: you can probably add some ssl cert bundle building or disable cert verification or all lp users will whine11:02
bialixmgz: re 599860: I can do a simple workaround by catching Unicode error there and try stringify11:02
mgzhm, you mean exposing the conf option as a checkbox or something vila?11:02
vilajelmer: around ?11:02
vilamgz: I mean if we don't provide a valid bundle all https connections will fail11:03
mgzbialix: for that bug, it's less about fixing the symptom and more making sure that once the core bzr stuff changes explorer doesn't break11:03
vilaso either we provide it or we set the ssl.cert_reqs option default value to none11:04
bialixmgz, vila: http://wiki.bazaar.canonical.com/BzrWin32Installer#id2011:04
bialixout-of-date a bit, but that's what I did back in 200811:05
mgzthanks bialix11:05
vilathanks bialix11:05
vilastereo effect :)11:06
bialixwe should check whether that still work with bzr.dev exe building11:06
vilamgz: doxxx may need to do the same for the osx installers11:06
bialixnice stereo11:06
vilamgz: I'm working on bug #920455 so I need to know what path to use11:07
ubot5Launchpad bug 920455 in Bazaar "ssl cert verification needs better defaults for all supported platforms" [High,Confirmed] https://launchpad.net/bugs/92045511:07
vilamgz, bialix: I probably should reuse bzrlib.transport.http.ca_bundle.py for that (and for now)11:08
bialixvila: at least that used to work11:09
jelmervila: yup11:09
vilajelmer: (requeing ;)  seen my messages about quilt for udd ?11:10
jelmervila: scrolling up now..11:11
bialixmgz: I'd be happy to prepare qbzr/explorer for bzr 2.5 with you.11:14
jelmervila: you can use quilt-smart-merge=False to disable it11:16
vilajelmer: in the package itself I presume ? (given the name with - instead of _ ;)11:17
jelmervila: yes11:17
vilajelmer: won't work for jubany11:17
vilajelmer: but keep reading, quilt has been deployed I don't need to disable at a global level anymore11:17
jelmervila: given it's a package dependency, I'm not sure if it's really necesary to cope better with quilt being missing as it should always be there11:17
mgzbialix: that would be great11:17
vilajelmer: dependency of bzr-builddeb ?11:18
jelmervila: yes11:18
vilaof course, I agree then11:18
vilajelmer: will teach me to deploy unpackaged packages ;)11:19
jelmerheh, indeed ;)11:20
mgzbialix: hm, mind if I just cherrypick the treewidget fix to 0.21 rather than rebasing? having the first branch landed and not wanting the last one there makes things a bit more confusing11:25
mgzwill add the news after the cherrypick11:25
bialixmgz: anything is fine for me11:30
mgzdoing that now, will propose so the diff passes by you11:32
vilaRiddell: cjwatson just mentioned he did sync pristine-tar 1.17 in precise ;)11:34
mgzdammit, this feature moving is just annoying me now11:35
vilamgz: rather than two months ago you mean ? :-D11:36
Riddellvila: lovelyness11:37
mgzI should update to latest beta, that backward import I think11:39
bialixmgz: feature moving?11:39
mgzha! now on 2.5b5 and... different error ;_;11:41
mgzplugins/qbzr/lib/tests/test_spellcheck.py", line 25, in <module> class _PyEnchantFeature(Feature):11:41
mgzTypeError: Error when calling the metaclass bases __init__() takes at least 5 arguments (4 given)11:42
mgzI should just get a 2.4 install so I'm running the right versions against each other11:42
mgzokay, tests pass on 0.2111:44
* mgz reverts to .po11:45
* vila out for lunch11:48
mgzga, I hate news12:25
mgzadded it to a released section, needed to make a new one12:25
jelmervila: hmm, I'm just looking at some test failures in debian12:55
jelmerit's not a problem if read() in bzrlib/tests/test_test_server.py line 77 raises a 104 error (connection reset by peer) ?12:56
bialixmgz: thanks12:56
mgzbialix: I think I got all the bits right in the end, but yell if you see anything I messed up12:57
vilamgz: hehe, welocme to the club ;)13:01
vilajelmer: let me have a look13:01
vilajelmer: trunk ?13:01
vilajelmer: ha, *that* read :-/13:02
* bialix just re-read the description of qbzr on lp: "A simple Qt cross-platform frontend for some of Bazaar commands" and thought how many of that is still true? simple? some commands?13:03
vilajelmer: my take here is that we should go back to where this code was introduced and split new tests instead of messing with the existing ones13:03
vilajelmer: the modifications were based on an assumption that has been proved wrong repeatedly and their intent has never been clear to me13:04
mgzhow do you do that neat tracked-in-0.22 series thing in launchpad?13:05
vilajelmer: add the fact that these tests tend to succeed for many known (and some more unknown) races to blurry the area...13:05
vilajelmer: but which tests do you see failing ?13:06
bialixmgz: click on Target to series13:06
bialixmgz: select trunk and desired series13:07
Riddellbialix: and it should11:27 < Tm_T> dpkg: error processing /var/cache/apt/archives/plasma-dataengines-addons_4%3a4.8.0-0ubuntu1~oneiric1~ppa1_amd64.deb (--unpack): trying to  overwrite '/usr/share/kde4/services/plasma-engine-kdeobservatory.desktop', which is also in package plasma-widget-kdeobservatory  4:4.7.4-0ubuntu0.113:07
Riddelloh sorry13:07
mgzah ah, trunk13:07
Riddellbialix: and it shouldn't mention Qt, that's an implementation detail13:07
bialixRiddell: I'm sorry I'm not quite understand13:07
Riddellbialix: "A simple Qt cross-platform.." Qt can be removed from there, it doesn't matter what widget toolkit you use what matters is the end result13:08
mgz...pasted the wrong thing Riddell?13:08
bialixRiddell: right, that's true, it's simple implementation detail13:09
Riddellmgz: yes, I shall now remove my stupid thinkpad touchpad with a stanley knife13:09
Riddellbialix: how about "A user friendly GUI to Bazaar commands and guided usage" ?13:09
bialixmgz, vila: I'd like to commit my workaround for bug https://bugs.launchpad.net/qbzr/+bug/91234413:10
ubot5Launchpad bug 912344 in Bazaar "Bazaar explorer crashes when in browse and I try to show annotate on a file" [High,Confirmed]13:10
bialixalthough that's clearly bug in bzrlib, but we can workaround it in qbzr itself13:11
mgzI'm still not used to middle-click paste, I'm alwasy trying to open a new tab but end up in a search for whatever's on (one of) my clipboard at the time13:11
bialixRiddell: good. But what do you mean by "guided usage"?13:12
bialixRiddell: but I think your description is better suited to Bazaar Explorer13:13
mgzsomething along the lines of Riddell's suggestion would work for qbzr13:14
vilabialix: workaround seems fine13:14
Riddellbialix: oh right maybe that's what I'm thinking of13:14
mgzjust need to get across the point that it's guis for individual commands13:14
jelmervila: test_handle_request_closes_if_it_doesnt_process is failing13:14
Riddellbialix: then maybe "A user friendly cross-platform GUI to Bazaar features"13:14
vilajelmer: yup, known but rare spurious one13:14
bialixRiddell: !13:15
jelmervila: is there any reason getting a 104 error there is invalid?13:15
Riddellbialix: what's that ment to mean? :)13:15
bialixRiddell: that mean: yes, that's good one13:16
Riddelloh nice, go me :)13:16
vilajelmer: long story short: can't say. I was against closing in handle_request as it broke the known working model...13:17
bialix"go me"?13:17
mgzbialix: while I have you, on an unrelated issue, are you aware of anything in bzr-explorer that can cause it to open lots and lots of connections to the same server?13:18
vilajelmer: *why* it breaks (as revealed by this failure) is unknown to me13:18
mgzbialix: Riddell is pleased you liked his suggestion13:18
bialixmgz: I think that's because we have separate codebase for inspecting working tree status + working tree widget + what's is not submitted to the parent branch13:19
mgzbut that's still only two or so connections right?13:20
bialixmgz: my own opinion on this is that there should be a dedicated process to do all bzr heavy-lifting, while bzr-explorer itself should be a thin frontend, like a browser13:20
bialixmgz: I've once described  my thoughts on this topic back in April or May 2011, but I haven't had enough spare time to implement this wild and heavy idea13:21
bialixmgz: that too many connection error also could be triggered in qlog alone13:22
mgzha, yes... what causes the error to be raised rather than just opening lots of connections though?13:22
bialixmgz: bzr-explorer re-fresh status from time to time, so I suppose it re-open the branch many times13:22
bialixconnection just is not re-used13:23
bialixmgz: yes, the branch should be reopened on refresh, otherwise in some corner cases we lost track of actual changes13:24
bialixlike switching light checkout to another branch behind GUI13:24
bialixwould you like me to find my old rant?13:25
mgzbialix: that would be nice, doesn't seem to be on the bzr list?13:28
bialixmgz: would you like to be listed in AUTHORS.txt list of qbzr with your full name, rather than [gz] ?13:29
bialixmgz: I found it, May 3, 2011. I see it in qbzr ML, should be in bzr ML as well13:30
mgzbialix: I'm fine with either13:31
vilabialix: he won't tell you but I'm sure he prefers [gz] ;)13:32
bialixvila: that was my suspicious too13:32
bialixmgz: https://lists.ubuntu.com/archives/bazaar/2011q2/072377.html13:34
mgzthanks bialix13:37
bialixmgz: actually, I still think it could help, but in the same time it will require too much efforts13:37
bialixmaybe re-using part of TortoiseBzr, but it too tight coupled to windows13:39
mgzwell, we can leave till after 2.5 at least13:39
bialixmaybe using python standard (in 2.6) processing library would help (a lot)13:40
bialixI meant "processing"13:41
bialixor it's called now "multiprocessing"13:41
bialixmgz: do you have a favorite tree?13:51
bialixwe need codename for 0.2213:51
vilabonzai ? :-D13:52
bialixit's not a specie, it's a form13:53
vilarats, for once the tyop was also a pun ;)13:53
mgzhm.13:53
mgzalder?13:54
bialixwrong word13:54
bialixsold!13:54
mgzapparently the one round here is alnus glutinosa or black alder.13:56
bialixnice13:58
mgzheh, and on the wikipedia page vila, it has a 'Bonsai' section13:58
vilalol, I'm on the wiki page but not there yet ;)13:58
bialixthis summer I saw exhibition of bonsai trees, it was nice. no alder unfortunately13:59
mgzbialix: after a qbzr (or bzr-explorer) release, who does the packaging for debian/ubuntu?14:08
bialixnot me14:08
bialixI'm not sure who make packages now14:09
bialixwho does the packaging for other bzr products?14:10
mgzsomeone who's name begins with J and ends with elmer does bzr itself,14:11
mgzperhaps I can twist his arm into doing or helping me with qbzr and explorer, if he doesn't already.14:12
mgzright, lunch time for me.14:12
LarstiQmgz: http://packages.qa.debian.org/q/qbzr.html points at http://packages.debian.org/changelogs/pool/main/q/qbzr/current/changelog which says mister Elmer and Andrew Starr-Bochicchio14:13
LarstiQlately at least14:14
smoseris this a known error:14:43
smoser$ bzr branch lp:~orchestra/orchestra/odev14:43
smoserYou have not informed bzr of your Launchpad ID, and you must do this to14:43
smoserwrite to Launchpad or access private data.  See "bzr help launchpad-login".14:43
smoserbzr: ERROR: Target directory "" already exists.14:43
smoserwhy would target directory be ""14:43
smoser?14:43
vilasmoser: sounds familiar but only for quite recent bzr versions, I think jelmer already fixed it14:46
smoserthis is precise "right now"14:46
smoser$ dpkg-query --show bzr14:47
smoserbzr     2.5.0~beta5-3ubuntu114:47
=== yofel_ is now known as yofel
vilajelmer: can you confirm ? ^14:55
mgzLarstiQ: thanks, now do I need to bug you to propose your pypy test fixes? :)15:01
jelmervila: no, that's not actually fixed yet but it's on my radar15:02
=== jml is now known as jml`
=== jml` is now known as jml
=== marienz is now known as 5EXAA07MX
=== 5EXAA07MX is now known as marienz
=== SpamapS_ is now known as SpamapS
=== james_w` is now known as james_w
=== deryck is now known as deryck[lunch]
* mgz drives jelmer nuts with non-review review comments17:37
* jelmer_ drives jelmer nuts by not being entirely awake today.17:38
* vila throws nuts17:39
mgzmeh, not a list admin, can't approve my message with too-large attachments17:40
vilause smaller attachments ;)17:40
mgzthat means downloading, compressing, and reattaching, but yeah, is probably sanest17:40
vilayou really need to send such a big attachment ?17:41
LarstiQthe limit isn't that high iirc17:43
vilawhich list ?17:43
LarstiQmgz: no, I should have some time today17:43
LarstiQmgz: between work/study/renovating there isn't much to go around17:44
vilain the good old days, merge proposals came with their associated bundle... and could be quite large17:44
LarstiQvila: hmm, true17:44
mgzinternal one, but 6MB was pretty large17:45
mgzLarstiQ: don't let me stop your carpentry17:46
LarstiQmgz: no worries, I need a break from it :)17:47
=== beuno_ is now known as beuno
=== deryck[lunch] is now known as deryck
wgzouch, a ValueError on missing certs dir? that's not nice18:56
fullermdHey, it's not making a value judgement.19:05
fullermdOr, wait, I guess it is...19:05
LarstiQwgz: at the end of bzrlib.hashcache.HashCache.read there is a comment we should use try/finally, is there a reason we're not doing so?19:08
* wgz checks19:21
wgzI think just that it was a pain to reindent everything and I didn't hit any cases that actually threw on the intervening lines19:22
wgztechnically things could, but only with broken files I think19:23
lifelesswell, that never happens :P19:25
wgz:)19:27
wgzif the error results in bzr exiting anyway, closing the file you're reading from promptly is less important19:28
vilawgz: Straight ValueError ? Or ConfigValueError  (the later should give proper info to the user) ? Do we need blackbox test for that maybe ?19:29
* vila off19:29
wgzvila: you changed one of them in your ssl defaults branch anyway19:31
vilawgz: right, IIRC I *removed* one possible cause of ValueError leaving only a ConfigValueError but imbw19:32
wgzdammit, I just tried to use vi on this box19:34
wgzI clearly can't segment my life properly.19:34
lifelesswgz: bzr is used as a library too though19:47
LarstiQwgz: what I have so far, minus tar export changes, is at lp:~larstiq/bzr/bzr-pypy19:47
lifelesswgz: so 'exit this call' != 'exit the process'19:47
LarstiQwgz: there is more work in a similar vein to do still, so not sure about proposing it for merging19:48
lifelessLarstiQ: does it work?19:50
lifelessLarstiQ: is it in the right direction?19:50
LarstiQlifeless: yes, no NEWS entry written yet19:50
lifelessdo that and merge IMNSHO19:51
LarstiQlifeless: ok. While I think on how to formulate that entry I can squash more test failures :)19:52
wgzyeah, I agree, am always a propose-first news-last person myself19:53
wgzwhen I remember news that is.19:53
=== _thumper_ is now known as thumper
=== kirkland` is now known as kirkland

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!