=== JanC_ is now known as JanC [00:08] ha nm, it's a bug in the perl [00:08] sigh [00:11] can anyone see why dpkg-parsechangelog says this is an invalid date? [00:11] Tue, 17 May 2008 10:93:55 -0500 [00:11] lol, just saw it [00:11] 10:93 [00:31] i wonder how that happened? fat-fingered a manual changelog edit? [00:31] smoser: --^ so we could do a parent override for that too [00:37] nacc, shoot. i must have looked at those wrong. sorry. the pcre3 and iscsitarget. [00:37] ohwell [00:40] smoser: it's fine [00:40] the iscsitarget one is fun [00:40] we're stripping the string we get back [00:40] so ' ' which would parse [00:40] gets turned into '' which does not :) [00:41] need to make usd [00:41] smoser: yep that's on my todo for tmrw [00:41] and convert clone to python [00:41] yep [00:41] i have one already going for that [00:41] a branch i mean [00:41] and for merge [00:41] that's my last main coding change i konw i'd like to be done with for 1.0 [00:41] beyond these bugfxies [00:42] the removal of dl-cache... [00:42] hrm? [00:42] it only gets set to true if provided on cmdline [00:43] 57cc02fb81675a2511400ff52f4a21c2a3f1100a [00:43] you didnt need to do that [00:43] yes i understood [00:43] or i dont understand. [00:43] did you read the commit message? [00:43] the cache by default (when not passed as an argument) is in .git [00:43] which will get deleted presuming --no-clean is not passed [00:43] but dl_cache is not set [00:43] workdir is [00:43] dl_cache will still be None [00:44] if dl_cache is None: [00:44] workdir = os.path.join(local_repo.git_dir, CACHE_PATH) [00:44] right [00:44] why does it matter if dl_cache is none or not? [00:44] we do not set dl_cache [00:44] if dl_cache: [00:44] remove stuff [00:44] er.. [00:44] wait [00:45] smoser: it's getting late here (and i've been starting at this all day), it's gotta be even later for you! :) [00:45] smoser: yeah, i think what was there before was confusing and a change in behavior -- i think we can do it, but let's do it a distinct change, if that's ok with you [00:45] oh. you're right. never mind. [00:45] smoser: this way, the cache is preserved if the directory is preserved [00:45] which i think is the most correct default [00:46] i suppose, yeah. thats fine. [00:48] nacc, elinks: [00:48] 11/02/2016 23:53:51 - DEBUG:stderr: gbp:error: The orig tarball contains .git metadata - giving up. [00:48] http://paste.ubuntu.com/23418680/ [00:49] and docker.io: http://paste.ubuntu.com/23418681/ [00:49] those are new failures from pristine-tar [00:49] i'll let you go, and i'm gonna go too [00:50] but basically, it seems to be going really well, this clean restart with the --dl-cache [00:50] 157 total. 16 working. 139 pass. 2 fail. [01:01] smoser: right, import-orig won't like .git-ified upstreams [01:01] which aren't supposed to keep their .git, but some do [01:02] smoser: let's resolve what we're going to do about the orig stuff, because i'm not going to spend time on fixing those if we're not going to merge that before 1.0 [01:02] right. [01:02] i think either way the other bugs take priority and we'd need to add a flag like --no-pristine-tar or something [01:02] maybe even default to that, dunno [01:02] smoser: in any case, thanks for your patience and insight, i'm done for the day too -- talk to you tomorrow! [01:03] you must have moved some stuff arodn for usd-build [01:03] before you committed [01:04] it doesn't compile right now [01:04] but tomorrow it will [01:04] good night [01:04] and Go cubs Go! === g2[CUBS-ATL] is now known as g2 [05:34] nacc: sounds like debian-devel [06:37] Ok, pulseaudio si sitting in zesty-proposed, because pulseaudio-equalizer requires packages from universe, yet it was NEWed into main. Can I do anything to help demote it to universe? [08:21] pitti: sudo mount-image-callback -d install_disk.img chroot _MOUNTPOINT_ /bin/bash [08:21] --system-mounts --system-resolvconf [08:40] http://paste.ubuntu.com/23419821/ [08:43] Could someone approve the trusty bug nomination on https://bugs.launchpad.net/ubuntu/+source/bash/+bug/1422795? I suspect that's why it's not shown on http://reqorts.qa.ubuntu.com/reports/sponsoring/index.html [08:43] Launchpad bug 1422795 in bash (Ubuntu) "bash crashes often if inputrc contains revert-all-at-newline" [High,Fix released] [08:44] ANYONER WITH CENTOS OR UBUNTU ROOT [08:44] ANYONER WITH CENTOS OR UBUNTU ROOT [08:44] ANYONER WITH CENTOS OR UBUNTU ROOT [08:50] hammed, no spam [08:57] ya [08:57] i need ubuntu [08:57] root [08:58] hammed: type sudo command-you-want-to-run-as-root [08:59] hammed: (just the same as on OS X an other modern systems) [09:00] root [09:00] where [09:01] hammed: what are you trying to do, that "needs root"? Perhaps we can help you better, if we understand [09:44] mardy: any chance you can build that package as an sru for yakkety too? [09:45] Laney: I think it has to land on zesty first [09:45] mardy: can be at the same time [09:46] but right now I want one to test [09:47] maybe the zesty binaries are installable [09:47] Laney: ok, but I think you can use the xenial version then, I would expect it to work fine on yakkety [09:48] Laney: As you had been quite helpful on the last trusty SRU I needed (thanks for that!)... Do you happen to know the best way currently to get an SRU to a package in main uploaded? [09:50] hey pkern [09:50] is it in the sponsor queue? [09:55] tjaalton: please could you complete the MIR issue #1547490 [09:57] i need just only ip user plus pass not [09:57] Laney: I think https://bugs.launchpad.net/ubuntu/+source/bash/+bug/1422795 isn't because it needs a coredev to approve the nomination or something. [09:57] Launchpad bug 1422795 in bash (Ubuntu) "bash crashes often if inputrc contains revert-all-at-newline" [High,Fix released] [09:58] Laney: ubuntu-sponsors is subscribed since somewhen in 2015. [09:58] pkern: ah, could be - let me do that [09:58] Maybe it would be a good idea to fix the queue to show bugs with pending nominations [10:01] Laney: Thanks! I was somewhat surprised that it didn't. Or that nominations to packages in main can't be done with usual bugcontrol privs. (Which would be the other recourse here.) [10:01] I think you have to be a "driver" for that - bug control just lets you nominate in the first place [10:01] IIRC [10:02] Ah, right. I think MOTUs are drivers for universe, though. (Which makes sense, so well, meh, maybe.) [10:04] driver *or* uploader [10:04] AIUI [10:12] doko: what's missing from -libinput? [10:12] tjaalton: the MIR [10:15] hmm ok === sil2100_ is now known as sil2100 === hikiko is now known as hikiko|ln === alan_g is now known as alan_g|lunch [13:00] hmm [13:00] latest update broke some of the indicators for me [13:01] https://www.dropbox.com/s/lx69qjcc25qnmnb/2016-11-03_14-00-30.png?dl=0 [13:03] andrewsh: i think you want #ubuntu instead. it's the support channel [13:04] well, I don't think so [13:04] I'm pretty sure I don't need support, but I need to report a bug instead === hikiko|ln is now known as hikiko [13:17] mardy: I get the notification again instead of the webview with the package from the silo [13:17] thanks! [13:17] Laney: \o/ [13:21] andrewsh, if you report a bug include some useful details like the version of ubuntu you are using and what you updated and what indicators are buggy and in which way [13:22] andrewsh: ubuntu-bug for each indicator [13:22] no [13:22] that is buggy [13:22] well, if it's an issue in the lib no point opening a bug on each indicator using it [13:23] would be nice to describe what is buggy to start [13:24] well the "broken" indicators look like they are application indicators anyway, not standard system indicators [13:24] how do you know? [13:24] he didn't even say which ones [13:25] 09:00 < andrewsh> https://www.dropbox.com/s/lx69qjcc25qnmnb/2016-11-03_14-00-30.png?dl=0 [13:25] because of their position on the panel [13:25] oh, weird url is a screenshot [13:25] could be bug #1635625 [13:25] bug 1635625 in unity (Ubuntu) "Some indicator icons are missing after unlocking the screen" [High,Confirmed] https://launchpad.net/bugs/1635625 [13:25] which is new in 16.10 [13:25] but not due to a recent update [13:25] so that doesn't really match the IRC description [13:26] yeah, and nothing has really changed in the library or indicator-application recently either [13:32] seb128: the indicators are from skype and owncloud === alan_g|lunch is now known as alan_g [14:28] nacc, smoser: let me know when you're both available to talk about the importer [14:29] rbasak, ok. [14:30] i'm fine whenever. just when the west coast wakes up. [14:30] OK [14:30] * rbasak has a hard stop today [14:36] cpaelzer: around? Did you do anything to your branch after the dovecot merge? It would do with tagging and pushing, and then I can run the importer to bring it up to date with your work. Maybe this isn't documented? [14:39] rbasak: I'm here [14:40] rbasak: I haven't touched it since the upload with the modifications to the MP that you reviewed [14:40] rbasak: but I might miss the point of your question [14:40] rbasak: since I'm curious - short HO to sync on that? [14:40] cpaelzer: please could you tag the thing that you uploaded? Should be upload/ (with substitutions if needed) [14:41] cpaelzer: sure! [14:41] rbasak: yeah in the past the uploader did that, since that now is me yeah I should add the tag [14:42] rbasak: I'm in the team HO that we have later [14:42] Oh === cmagina_ is now known as cmagina [14:56] cyphermox: I noticed you rebuilt devscripts at the start of the yakkety cycle; could you do the same for zesty now so that it picks up zesty as the current dev release? [14:56] sure [14:56] but I thought this already worked [14:57] or not [14:59] jbicha: I might as well do the merge at the same time, so it will take a little while [14:59] cyphermox: that's fine, thanks [15:05] rbasak: i'm around now [15:05] smoser: --^ [15:06] nacc, smoser: usual team hangout? [15:06] k. yeah. [15:06] rbasak: smoser: sure [15:12] Whoa, I spent an hour on cleaning up a bugs now, I got 19. [15:13] There are still 446 in LP, if anyone else wants to play. [15:13] :) [15:18] mardy: oh no! [15:18] got a webview out of the blue [15:20] lots of signon stuff in syslog /me attaches [15:20] Laney: and you still have the signon-ui from that silo? [15:20] yeah [15:20] didn't restart [15:21] I notice that the title is "Web authentication" instead of "Web authentication for Google" [15:21] mmm [15:21] Laney: and what are the contents? Still google? [15:22] yep [15:22] sec, just attaching [15:23] jbicha: done. [15:23] ahh, I forgot to remove URLs and stuff [15:23] mardy: it's there, no massive rush on investigating though [15:24] Laney: thanks, I'll have a quick look now just to see if I need to ask you for some more [15:31] cyphermox: missing -v for devscripts merge [15:32] Laney: could it be that you removed some lines coming from com.nokia.singlesignonui[11563] ? [15:32] mardy: not removed, but I maybe missed them when copying (just selected a range) [15:32] lemme see [15:35] mardy: There's stuff from earlier on when I logged in [15:37] Laney: no, that's not important [15:37] Laney: I guess I need to add some debugging statements here and there [15:40] Laney: eds is really spamming signond with 28 requests in a second (and apparently, for a single account), maybe it's triggering some race condition [15:41] Laney: I see that the first request is going well, reusing the previously stored token without invoking signon-ui [15:48] doko: oh, you're right === jamespag` is now known as jamespage [17:41] smoser: i don't know what happened to your fetchmail import, it worked fine here -- let me finish testing the fixes for LP: #1638614 andi'll push to master [17:41] Launchpad bug 1638614 in usd-importer "linux-base, module-init-tools fail import: AttributeError: 'NoneType' object has no attribute 'group'" [Medium,In progress] https://launchpad.net/bugs/1638614 [17:41] nacc, but it didnt stack trace [17:42] smoser: yeah, i have no idea -- did the system kill it? [17:43] smoser: it also doesn't show any errors in that paste [17:44] dont think so [17:44] well, it exted non-zero was all [17:44] but yeah, it was odd to me too [17:46] smoser: right, but we always exit with 1 on failure [17:46] iirc [17:46] so if it was a different exit code [17:46] then maybe it was sigkill'd? [17:46] i just had one die to too many open files [17:47] which i'm not sure how htat happened yet [17:47] i think system is kind of just really not happy [17:48] so possible that something happened [17:48] yeah, i'm not sure how to debug it at this point [17:49] smoser: ok, so git doesn't like '' as an author :) [17:49] smoser: suggestions of how to resolve this? [17:50] i could i think walk the lp package_maintainer_link maybe [17:51] nacc, yeah, not worry about the fail if not reproducible http://paste.ubuntu.com/23421781/ is my dmesg. that thing does look scary [17:52] wrt git not liking ''... [17:52] if author == "": [17:52] commit_author = "NotAvailable" [17:52] or somethign to that affect ? [17:53] right, but then we are putting that in the commit hash :) [17:53] and that's not obtainable without the importer [17:53] i'm trying to think if there's anything better we can do [17:55] meh. [17:55] lots of things are not obtainable without the importer. [17:56] i guess that's true [17:56] so i do think launchpad figures this out [18:02] testing again [19:24] smoser: ok, was able to fallback to launchpad data for the name, at least; if the email fails to parse its a hard fail as you need to be logged into lp for that data [19:24] but fixed the code to fail cleaner in that case, at least [19:37] smoser: so i think there migth be some underlying leak with the commit_msg_file thing [19:37] smoser: i just checked with lsof and even though the files are listed as deleted, they are showing as still open [19:37] any idea why that might be? [19:50] nacc: (also some people have their email addresses hidden in LP, so you need a fallback anyway) [19:50] cjwatson: ack [19:51] smoser: we probably need to handle that properly at some point (email failing to parse fully) [19:51] smoser: i wonder if we could just insert or something [19:52] hm.. nacc reading for commit_msg_fiel comment [19:53] smoser: thanks, i don't see it yet, unless it's a side-effect of the with and unlink [19:53] smoser: is it possible the run is holding on to a reference somehow? [19:53] nah. [19:53] so... [19:54] tempfile.mkstemp() returns a 'os' filehandle and the file name [19:54] we open the file name on our own (with open ...) and close that [19:54] but i suspect possibly the os file handle needs to be closed [19:55] ah, so it's the first filehandle still being open [19:55] that would make sense [19:55] os.close(_) [19:55] essentilally [19:55] perhaps you should use os.fdopen instead [19:55] (closing a file object created by fdopen closes the underlying file descriptor too, which is probably what you want here) [19:56] so you wouldn't need a separate os.close if you did that [19:56] (I have not read the code here, I just know fdopen) [19:57] possibly, yeah. [19:57] or else just use a higher-level thing like tempfile.TemporaryFile or tempfile.NamedTemporaryFile rather than messing around with mkstemp [19:58] which is implemented with mkstemp + fdopen anyway, but why bother doing it yourself when you don't have to [19:58] smoser: is it not possible to do fp, _ = tempfile.mkstemp() [19:58] fp.write(msg) [19:58] fp.close()? [19:58] or as cjwatson said [19:58] nacc: that's: [19:58] you need os.write() [19:58] with tempfile.TemporaryFile() as fp: [19:58] to that thing. as its hot a file. [19:58] fp.write(msg) [19:59] (though obviously you would need to do something more as a temporary file that immediately vanishes isn't very useful) [19:59] ah right [19:59] we want a name to it also . which is why i used mkstemp [19:59] smoser: then use NamedTemporaryFile. [19:59] you're fighting the stdlib unnecessarily here. [19:59] smoser: right, so there is an issue here, can you spin up a fix? :) [20:00] smoser: it seems fatal only for libvirt so far :) [20:00] sure. i'm not sure why i didnt use it. but sure. [20:00] smoser: thanks! [20:01] you may need delete=False and a later unlink, or else do fp.flush() and put the stuff that uses the tempfile inside the with statement [20:02] anyone have a suggestion on how to commit the following: https://launchpad.net/ubuntu/+source/iscsitarget/0.4.15+svn148-2.1ubuntu1 [20:02] note the date is wrong [20:02] as in, invalid -- so should i fallback to the publishing information? [20:13] nacc, i suspect http://paste.ubuntu.com/23422293/ fixes [20:13] smoser: looks reasonable [20:13] actually. [20:15] http://paste.ubuntu.com/23422302/ [20:16] makes sense [20:17] you want a MP? [20:17] smoser: i can commit it here if that's ok by you [20:17] I think I'd just put the run() and whatever's below it inside the with NamedTemporaryFile block in that case and not have to bother with delete=False [20:17] assuming it's small [20:17] yeah, that seems much cleaner [20:17] thats fine. [20:17] smoser: ack, commiting now [20:18] cjwatson, i probably need the flush, right ? [20:18] yeah [20:25] nacc, that "only" leaks one file handle per git commit though. [20:26] so its not great, but this glibc import i have running for 18 hours now only has 417 open files. [20:27] smoser: committed & pushed [20:27] cjwatson: thanks! [20:27] yeah, it's not a huge deal, just an oddity when i was hopeing to use `lsof -p | grep /tmp` to quickly see where it was running without having passed --verbose [20:30] smoser: thoughts on what to use as a fallback for the author date if it is unparseable? i am adding a strptime check on the extracted value from the changelog, but what should i use if it's bad? published date if possible? and if neither is present ... [20:31] if neither is present i suggest my birthday [20:31] i dont know. publish date shoudl be present always, shoudlnt' it ? [20:31] it's not always in my expereicne [20:32] i can't recall if that's because of superseded or deleted values [20:32] let me see if the underlying pacakge upload date always is [20:52] smoser: do you think it would be trivial to disconnect the run logging from the verbosity? that is, it'd be really nice to run with --verbose and get just the tracing we have from the importer code, not the shell commands being run. To get those, maybe -vv or something? [20:53] smoser: when debugging one change that will fix an import, i really can't read the output anymore :) [20:53] smoser: and i get all the cache hits too [20:54] its not terribly hard. really we just want to set up different loging i think (at lesat thats one way to do it). [20:54] -v = info [20:54] -vv = debug [20:54] -vvv = loud [20:54] right [20:55] but run and co use logging.debug [20:55] as does the importer [20:55] so just changing the loglevel globally won't work [20:55] i guess we could switch those around [20:56] yeah i think that's what we should do [20:56] well, we have to add a 'loud' [20:56] which is just [20:56] logging.log(logging.DEBUG - 10, msg) [20:56] or logging.addLevelname [20:56] oh i didnt know this. [20:57] probably its and [20:57] not or [21:00] reading about it now [21:00] i think that would be more maintainable [21:02] err, right, it'd be something like logging.addLevelName(logging.DEBUG - 5) [21:02] as DEBUG is 10 and 0 is NOTSET [21:02] i'll play with that [21:02] sure. ijust picked a number [21:03] i've been annoyned before of the "top heavy" ness of python logging. [21:03] yep [21:03] info , warn, error, critical [21:03] it doesn't lend itself to usage like we're wanting. [21:04] debug, debug2, debug3 [21:04] so cloud-init just gives you a firehose at debug [21:04] there really seems like there shoudl be a trace [21:05] yep [23:02] smoser: fudge, found a reason *not* to use a cache [23:02] smoser: https://launchpad.net/debian/+source/pcre3/1:8.35-3.2 [23:02] https://launchpad.net/debian/+source/pcre3/2:8.35-3.2 [23:02] note that the .dsc files are named the same [23:02] but are different [23:03] i believe pull-debian-source doesn't detect that [23:05] urp, they *should* be different [23:05] but they aren't === g2 is now known as g2[ATL]