=== wedgwood is now known as wedgwood_away [00:06] davechen1y: ping [00:09] ack [00:09] thumper: ack === davechen1y is now known as davecheney [00:09] davecheney: hangout? [00:09] yy [00:10] davecheney: have you got video working? [00:10] no video on this x220 [00:10] why? [00:10] that is a thinkpad right? [00:10] you can video, but bodgy sound (via n7) or good sound, and no video [00:10] davecheney: what about skype? [00:10] sure, lemme install it [00:10] i had al lthe tools on my mac [00:11] but it died while I was on vacation [00:11] :( [00:11] and a long series of sad stories and trips to the apple store have not resulted in it working again [00:11] I've tried to avoid apple hardware as much as possible [00:11] don't want to start down that track [00:11] I do know others that are almost 100% in to it [00:12] like popey and njpatel [00:12] i think avoiding it is a good idea [00:12] they use bcm wifi chips which are poorly supported [00:12] so I'm looking for a non-apple macbook air type equivalent [00:13] small, thin, metallic (robust), with good battery life [00:13] stay away from dells [00:13] they are all junk [00:13] the X1 isn't bad [00:13] lenvovo x1 [00:13] some of the samsung ultrabooks are quite nice [00:13] byut hard to get in au [00:13] oh, I forgot cheap (ish) [00:13] so i havne't seen many in the flesh [00:14] I've seen quite a few samsung ultrabooks here [00:14] jb hifi has them here [00:14] some are ok, some are just weird [00:14] stay away from sony stuff, that is actually wierd [00:14] weird [00:14] I had a sony laptop a while back, proprietary hardware, non-open drivers caused a pain [00:14] yeah, total crap shoot [00:15] asus make a lot of models [00:15] its going to have very very stock hardware inside [00:15] some of them are ok, but for me it is hard to shake their reputation as making cheap and nasty kit [02:02] thumper: it's lunch o clock now, afk for prob 40-60 mins to get some lunch for me and my misses [02:30] davecheney: ack [02:30] davecheney: let me know when you're back [02:46] * thumper afk for a bit [03:20] thumper: ping [03:20] hey, just otp [03:21] no wuks [03:23] kk , ready now [03:23] davecheney: skype again? [03:24] y [08:02] mornin' all [08:44] morning [08:47] hey all, william asked me to tell you there's a power cut in his area at present, in case you wonder where is he === dimitern is now known as dimitern_afk [08:50] dimitern_afk: Thx for the notification. And good morning btw. ;) [08:51] any estimates on when juju 2.0 will be out? [08:51] good job making so much progress on the Go port, all. Very exciting... [08:51] is v2.0 when the Go version will be the primary one that you encourage people to use? [09:01] elimisteve: the goal is to have the Go port the default version for 13.04 (Ubuntu Raring) [09:02] jam: great, thanks === dimitern_afk is now known as dimitern [10:17] TheMue: morning :) [10:17] dimitern: Hiya [10:33] i think it's high time to do something about keeping the channel topic in sync with what's happening in the project, so the people could get an idea quickly, like #launchpad channels [10:37] dimitern: oh, yes, good hint === edamato is now known as edamato-afk [11:14] fwereade: yo! [11:14] fwereade: power back then, i guess [11:14] fwereade: hey [11:15] rogpeppe, heyhey; and, yeah, a little while ago, but I forgot to open irc again ;0 [11:15] dimitern, heyhey [11:15] rogpeppe, I'm getting the rpc module tests timing out after 600s [11:15] fwereade: i *just* realised that i forgot to put in the timeouts you requested [11:15] rogpeppe, haha [11:15] fwereade: am currently editing rpc_test.go! [11:15] fwereade: (as of 5 mins ago...) [11:16] * fwereade gets to feel all smug and prophetic ;p [11:16] bah :-) [11:16] fwereade: do you want to discuss shortly https://codereview.appspot.com/7301085/ ? [11:19] dimitern, sgtm, I'll start a hangout [11:20] fwereade: cool [11:22] fwereade: actually, since it's almost standup time, can we do it around 13h? [11:22] allenap: poke [11:22] I'm just checking on https://code.launchpad.net/~allenap/juju-core/break-out-juju-store/+merge/142564 [11:23] and seeing if you need it landed, or if it needs updating, etc. [11:23] jam: Hello. [11:23] The last comment was only a few days ago, so I think I can just land it for you/ [11:23] ? [11:23] jam: I /think/ it's up to date, but I'll check. [11:23] dimitern, ah, yeah, sure -- might be eating then, but ping me when you're free and we'll figure something out [11:23] fwereade: sure, no rush [11:24] lunchtime [11:25] rogpeppe: Anything else on https://codereview.appspot.com/7313081/? [11:26] niemeyer: looking [11:27] niemeyer: i think zero is a reasonable length for a file, and i think think PutAll should work with all files [11:27] rogpeppe: It's not up to us [11:27] rogpeppe: It does work, though [11:28] rogpeppe: But we can't test that live [11:28] niemeyer: why not? [11:28] rogpeppe: Because it's not up to us whether an upload works with no content or not [11:28] jam: Yes, it's up to date. [11:29] niemeyer: if it doesn't work, we should return an error [11:30] rogpeppe: Amazon returns the error [11:30] niemeyer: if the file length is zero, we don't make any calls for amazon to return the error, no? [11:31] rogpeppe: Multipart uploads of files of length zero do not work [11:31] rogpeppe: I don't know what you're trying to achieve [11:31] niemeyer: i'd like to be able to iterate over all the files in a directory and call PutAll on each one, regardless of the file's size [11:31] jam: Forgot to say: yes, please, I'd be grateful if you could land it :) [11:32] niemeyer: presumably you're allowed buckets holding zero-length files? [11:32] rogpeppe: *multipart uploads*.. that's what this CL is about [11:32] rogpeppe: and I can understand your desire.. you should file a bug against S3. :-) [11:33] niemeyer: PutAll works with a file that's 1 byte long. i think it should work if it's 0 bytes long, and i don't think it's hard to make it do that. [11:34] niemeyer: perhaps it should use multipart only where necessary. [11:35] rogpeppe: PutAll is a method on Multi [11:36] niemeyer: ahem, good point :-) [11:40] niemeyer: well at the least, if multipart uploads of files of length zero don't work, we should return an error in that case. [11:40] rogpeppe: Sorry, we're going in circles [11:40] [11:31:00] rogpeppe: Multipart uploads of files of length zero do not work [11:41] rogpeppe: Amazon will say so.. I don't think there's any benefit in us reporting it early [11:41] rogpeppe: Amazon returns the error [11:41] niemeyer: ah, you get the error from Complete? [11:41] rogpeppe: It will fail in whatever way Amazon chooses to fail [11:42] rogpeppe: Just like everything else [11:43] rogpeppe: Look.. I can put an error result there if that makes you happier [11:43] niemeyer: ok, that seems reasonable - perhaps a doc comment rather than an error - amazon may fix it in the future, i suppose. [11:43] rogpeppe: It may [11:44] niemeyer: (failing when amazon chooses to fail seems reasonable, i mean) [11:44] rogpeppe: Cool [11:49] niemeyer: replied [11:49] rogpeppe: Actually, I just found a way to make it work [11:49] rogpeppe: I'll fix it [11:50] niemeyer: cool. the less edge cases, the better i like it :-) [11:50] rogpeppe: Amazon happily takes it if I upload a part with zero content [11:50] rogpeppe: It is an edge case still [11:50] niemeyer: only for goamz, not for the person using it, hopefully [11:50] rogpeppe: It is an edge case for the API [11:51] rogpeppe: People may not care in most sane cases, though === edamato-afk is now known as edamato [11:51] rogpeppe: The result of len(parts) with no content is 1 rather than 0 [11:52] niemeyer: yeah. it seems reasonable that a file should always have at least one part though. [11:52] niemeyer: even if it's zero-length. [11:52] rogpeppe: It's an edge case. [11:53] niemeyer: agreed. seems like the least-surprises way forward though. [11:54] rogpeppe: As I said, I'll fix it [11:54] niemeyer: cool, thanks. [12:00] rogpeppe: np, thanks for the review [12:01] In the future we can have an API in Bucket itself which chooses properly whether to upload a multipart or not based on size [12:01] But I'd rather fix the overall package API first [12:14] niemeyer: that sounds good [12:16] niemeyer: i thought i remembered that dave cheney made some fixes to gocheck so that c.Logf wasn't racy if called in multiple goroutines. am i imagining things? (i get lots of errors from the race detector, which would be nice not to see) [12:16] rogpeppe: Probably lying in the queue [12:16] rogpeppe: I'll have a look later [12:16] niemeyer: ok, cool. [12:17] niemeyer: i was initially slightly worried by my 5 data races :-) [12:18] rogpeppe: Yeah, quite misleading [12:20] fwereade: fairly trivial. https://codereview.appspot.com/7324046 [12:21] fwereade: and i'd like to know where your tests fail with this applied. [12:36] allenap: btw, I'm changing the signature of one of the testing/charm.go apis, which effects "store/branch_test.go" and "store/store_test.go". [12:39] rogpeppe: so I see the tests hanging as well, and it is nice to not have them hang, but doesn't that mean the test is actually still failing? [12:39] jam: yeah - it shouldn't happen, but tests pass for me. [12:41] jam: until i can reproduce the problem, i'm not going to be able to fix the bug [12:42] rogpeppe: http://paste.ubuntu.com/1643683/ [12:42] rogpeppe: are you running go 1.1 or some variant thereof? [12:42] I was running go1 (default on precise) but it turns out lbox doesn't work with it. [12:42] jam: yes - although i think i've run the tests on 1.0.2 [12:43] So I'm now running go-1.0.3 from ~gopher's ppa [12:43] jam: ah! [12:43] jam: this probably is highly indicative: cannot unmarshal null into Go value of type rpc_test.stringVal [12:43] rogpeppe: yeah, it says the same for the struct{} object. [12:45] jam: i guess i only ran the live tests against 1.0.2. [12:45] jam: because i see the same failure. [12:45] jam: thanks. [12:45] rogpeppe: yeah, prob just a json change between go std lib and how they handle nil [12:45] jam: yes, i think so [12:58] rogpeppe, http://paste.ubuntu.com/1643711/ [12:58] rogpeppe, ha, just like jm's [12:58] rogpeppe, sorry [12:58] fwereade: yeah, i'm on it. it's a go1.1 compatibility issue [12:58] fwereade: np [13:18] jam: Okay, once you've done that, I'll merge the relevant changes into juju-store. Or, do you want to land my branch before making your changes, and I'll fix-up juju-store? [13:32] allenap: it doesn't change your branch, it is just going to break juju-store [13:32] your branch and mine have both landed. [13:32] so you'll just want to touch juju-store [13:32] which is generally just removing a parameter. [13:32] jam: Okay, I'll look into that. Thanks for sorting this out! [13:33] allenap: lp:~jameinel/juju-core/no-symlinks is the branch, you can look at -r -2 [13:33] but it is basically just remove: "series" being passed. === TheRealMue is now known as TheMue [13:41] fwereade, rogpeppe: time for a first review? https://codereview.appspot.com/7323059 [14:01] TheMue: will look after lunch [14:01] * rogpeppe goes for lunch [14:01] rogpeppe: Great, thx. [14:01] rogpeppe: Enjoy. === wedgwood_away is now known as wedgwood [14:24] fwereade_: is it usual for the worker/uniter tests to take long? [14:24] dimitern, yes, afraid so [14:24] dimitern, I wasn't avle to find any particular hotspots [14:24] fwereade_: ok :) [14:28] TheMue, please remind me: how's public storage useful to the local provider? [14:29] TheMue, won't it implicitly upload tools anyway, rendering the whole thing redundant? [14:29] s/upload/"upload"/ [14:31] fwereade_: *** Test killed: ran too long. ?? [14:31] dimitern, ok, no, they shouldn't take that long [14:31] fwereade_: am I doing something wrong? [14:31] fwereade_: I just run go test ./... in worker/uniter [14:32] fwereade_: does it need internet access during the tests? [14:32] dimitern, no, shouldn't do, and that sounds sane (even if I always run from root -- go test ./worker/uniter/... -- but I doubt that's remotely relevant) [14:33] fwereade_: hmm, well I'll try running them in order going into each subdir to see which one is to blame [14:33] dimitern, I'm just running them myself, because I forget exactly how long they should take -- enough to be annoying, certainly, but nowhere near test timeout limit [14:33] fwereade_: ah, actually it shows - FAIL launchpad.net/juju-core/worker/uniter/charm 600.002s [14:34] fwereade_: all others are ok [14:34] dimitern, hmm, very interesting: try that one with -test.v -gocheck.vv [14:35] fwereade_: [LOG] 58.01524 JUJU mongod: error command line: unknown option sslOnNormalPorts [14:35] [LOG] 58.01531 JUJU mongod: use --help for help [14:35] that's almost immediately at the beginning [14:35] dimitern, what mongo build are you using? [14:35] fwereade_: and it just stays there [14:35] fwereade_: mongod -v? [14:36] dimitern, did you get the special version with ssl support? [14:36] fwereade_: git version: d1b43b61a5308c4ad0679d34b262c5af9d664267 [14:36] dimitern, it's in an ec2 bucket somewhere iirc [14:36] fwereade_: not sure, I might have installed it from the archive [14:36] dimitern, http://juju-dist.s3.amazonaws.com/tools/mongo-2.2.0-precise-amd64.tgz [14:36] fwereade_: it says 2.2.2 [14:37] dimitern, you need this build for SSL support, which is not optional [14:37] fwereade_: how about quantal? [14:37] I suspect there's one if you s/precise/quantal/ [14:37] fwereade_: same version? [14:37] dimitern, yeah, think so [14:38] dimitern, take a look at the bucket in a browser, I think it lists everything for you anyway [14:38] fwereade_: and I should purge the currently installed mongo as well? [14:38] dimitern, up to you, but make sure the tests find the ssl version [14:38] fwereade_: the bucket reports access denied on dir listing, but direct link works [14:38] dimitern, ah ok, maybe that changed [14:39] fwereade_: where should I put the extracted tarball? [14:41] fwereade_: anywhere in $PATH should work? [14:41] dimitern, yeah, should be fine [14:43] fwereade_: great, 10x - now the tests pass ok and fast [14:45] fwereade_: Sorry, phone [14:46] TheMue, np [14:47] fwereade_: Maybe oriented too much at the dummy environ. If I've seen right it also uses two different storage that are returned by Environ. [14:48] TheMue, in the doc we agreed to just return an empty storage [14:48] TheMue, and I'm confused by the way they appear to share a port..? [14:49] TheMue, if we can just do a dumb empty-on-list, err-on-get StorageReader I think that lets us drop all this confusing private/public stuff [14:49] fwereade_: The URL depends on the environ name and if it is started as private or public. And it is empty. [14:50] TheMue, it kinda looks like an awful lot of additional complexity to save two one-line methods on a new type [14:51] fwereade_: Please note it in the review. [14:51] TheMue, will do, just checking in case I was missing something obvious [14:52] fwereade_: To me the symetric usage of the storage has been more simple. [15:03] fwereade_: meeting? === teknico_ is now known as teknico [15:46] rogpeppe, a thought: is the set of available arches not the intersection of those available on the provider with those we can find tools for? [15:47] fwereade_: that seems a reasonable thought [15:47] rogpeppe, in which case I don't really think it's ok to require that someone recompile the client to use another cloud... [15:48] rogpeppe, additional exercise: how are we meant to discover those values from the CLI? I don;t think we can safely assume the existence of an environments.yaml [15:48] rogpeppe, sure, we won;t figure out how to drop it *soon* [15:48] rogpeppe, but still... [15:48] fwereade_: GUI clients won't have environments.yaml [15:49] rogpeppe, indeed [15:49] fwereade_: i didn't mean to suggest we should recompile the client to use another cloud [15:49] fwereade_: one mo, i need to interact in #juju-gui [15:50] rogpeppe, I admit, I put those words into your mouth, but I think it's a consequence of specifying the list of "available" architectures on the client [15:50] rogpeppe, np [16:18] fwereade_: Just proposed a change with no more separation between public and private but with using environs.EmptyStorage. [16:19] TheMue, awesome [16:19] TheMue, whoops, I never actually hit m [16:19] TheMue, I'll drop the redundant comments [16:22] TheMue, sent a couple, let me know what you think [16:23] fwereade_: yep, cheers [16:28] fwereade_: Yep, both makes sense, storage-port (maybe more settings for the storage backend will follow) and the comments. As you see I so far mostly panic, it's just an interface copy for quick compilation. ;) [16:29] TheMue, cool, thanks === wedgwood is now known as wedgwood_away === wedgwood_away is now known as wedgwood [17:30] rogpeppe, is trunk still expected to be broken? [17:30] fwereade_: yes. here's a branch to fix it: https://codereview.appspot.com/7324048 [17:30] fwereade_: sorry about the delay. [17:31] fwereade_: it took a little more than i expected. [17:31] rogpeppe, no worries, just syncing up [17:31] fwereade_: review appreciated [17:31] dimitern, jam, anyone else: ^ [17:31] rogpeppe: i'll take a look [17:31] dimitern: ta! [17:36] rogpeppe, looking good so far; you might be interested in https://codereview.appspot.com/7324049 [17:37] rogpeppe, LGTM [17:38] fwereade_: thanks [17:38] aaaand I'm done for now -- but I might pop back on later [17:38] gn all [17:42] rogpeppe: reviewed [17:42] dimitern: thanks! submitted. [17:42] fwereade_: gn [18:33] right, that's me for the day. first watcher test now failing in a useful way. [18:33] g'night all [19:57] morning [20:01] mramm: morning, you around now [20:01] ? [20:02] thumper: yep [20:02] I was just about to send an e-mail because I will be out most of the day tomorrow === wedgwood is now known as wedgwood_away [20:12] thumper, heyhey -- I have a grumpy daughter who might take a while to get to sleep, but hope to be able to swing by again later [20:12] fwereade_: ok, np === wedgwood_away is now known as wedgwood [20:36] "go install launchpad.net/juju-core/cloudinit: open /usr/lib/go/pkg/linux_386/launchpad.net/juju-core/cloudinit.a: no such file or directory" [20:37] that happens when I run "go get -v launchpad.net/juju-core/..." [20:38] hi benji [20:39] hi! [20:39] benji: I can't help, but just saying hi [20:39] heh [20:39] actually, that looks like a permission problem, I guess it should be "sudo go get [...]" [20:47] benji: well, should only need sudo if you don't have a go path set, no? [20:57] benji, thumper: definitely no need to use sudo [20:57] hi rogpeppe [20:57] rogpeppe: why are you still up? [20:57] benji: looks like you might not have GOPATH set [20:57] or online at least [20:57] thumper: i'm not :-) [20:57] thumper: honest [20:57] so this is your irc bot then? [20:57] yup [20:57] rog made me very clever [20:58] are you up for a chat at some stage? [20:58] hmm, GOPATH is set [20:58] benji: and exported? [20:58] thumper: definitely [20:58] yep [20:58] rogpeppe: when would be good? [20:58] bit of an issue with you being on the other side of the planet [20:58] thumper: when's good for you? [20:59] rogpeppe: ideally between 9 and 5 local :) [20:59] heh [20:59] thumper: what's your time zone? [20:59] UTC+13 [20:59] thumper: what time is it now for you/ [20:59] ? [20:59] 10am [20:59] well, either way "go get -v launchpad.net/juju-core/..." (no sudo) now works, I haven't touched GOPATH since it failed [21:00] thumper: 10am thursday? [21:00] yes [21:02] thumper: busy for probably an hour, but should be able to make half an hour or so after that, if you're around all today [21:02] thumper: i'll ping you [21:02] ok [21:18] thumper, ping [21:18] fwereade: hi [21:19] thumper, heyhey, I'm free for a chat if you like -- g+? [21:19] fwereade: g+ sounds good === hatch_ is now known as hatch [22:05] thumper: ping [22:05] rogpeppe: hey, just talking with fwereade now [22:05] thumper: i could join if you fancy it [22:06] thumper: depends what you're chatting about really [22:06] rogpeppe: propable one on one intially would be better [22:06] thumper: sounds good [22:07] thumper: unfortunately my window of opportunity is short tonight, so maybe we won't make it [22:07] rogpeppe: ok, perhaps tomorrow? [22:07] thumper: possibly, though unlikely. [22:07] thumper: evenings not usually good for me. [22:07] rogpeppe: I could get up early [22:07] 17:00 UTC [22:11] thumper: that sounds possible, yeah [22:12] thumper: i'll schedule it in... [22:12] ok [22:15] thumper: you have an invite :-) (means i might get a reminder...) [22:15] :) [22:16] thumper: if you get off chatting with william tonight, ping us anyway. i might still be around.