=== schwuk_away is now known as schwuk
=== anthonydillon is now known as antdillon
antdillonHi guys, is there anyone online from the juju gui team?11:46
rick_hantdillon: yep11:51
antdillonrick_h, Hey, thanks. Just a quick question about the browse section of jujucharms.com 11:52
rick_hantdillon: sure thing11:52
antdillonrick_h, Were linking directly to charms from the new juju.u.com website11:53
rick_hantdillon: ok11:53
antdillonrick_h, For example using this URL: https://jujucharms.com/precise/mongodb-15/11:53
antdillonrick_h, Is there a way to link that wont go out of date?11:53
rick_hantdillon: yep, so for now it needs to be -HEAD vs -1511:53
rick_hantdillon: in the future we hope to be able to drop that, but it's not complete yet11:54
antdillonrick_h, As in https://jujucharms.com/mongodb-15/11:54
rick_hantdillon: oh...ditch the precise...ugh no. We don't have anything for that and not sure it would work as most charms don't exist outside of precise atm11:54
antdillonrick_h, Because we dont want to link them to percise when saucy is out11:54
rick_hhttps://jujucharms.com/fullscreen/search/?series=raring&text= for instance is raring's charm list11:55
rick_hantdillon: yea sorry, we'll have to deal with updates when we know we want to update them. I don't think you'll want saucy links when saucy is out11:56
rick_hantdillon: you'll want to use https://jujucharms.com/precise/mongodb-HEAD11:56
antdillonrick_h, Ok cool, I'll leave it with precise in the url11:59
antdillonrick_h, Thanks for the help. Might be worth thinking about creating short links to the charms for sharing11:59
rick_hantdillon: yep, it's a bit complicated since you need to match your juju environment version to the charm version and such12:00
rick_hantdillon: but will bring it up on the call today and might be a good sprint topic in Oct12:00
antdillonrick_h, https://jujucharms.com/precise/mongodb-HEAD/ the tweet link tweets to https://jujucharms.com/precise/mongodb-15/12:00
rick_hantdillon: yea, we should fix that. -HEAD is relatively new12:01
antdillonrick_h, Cool thanks12:01
gary_posterhey antdillon.  charms are per series, and generally in Juju we care about the LTS charms not the dev/6 mo. release charms.  So what you raise is an interesting question but not pressing until we start thinking about T/14.0412:08
gary_poster(the next LTS IIRC)12:08
antdillongary_poster, Cool I understand12:20
antdillongary_poster, So should use -HEAD for all charms?12:20
gary_posterantdillon, yes, for now.12:22
bacgary_poster: i'm here today12:23
gary_posteryay bac!  100%, or?12:23
gary_posteras in feeling 100%12:23
bacno, but i'm faking it.  :)12:24
gary_posterbac, heh, ok, take it easy12:25
marcoceppihey guys, what does the charm use to serve the pages? Apache?12:31
gary_postermarcoceppi, atm, yes.  with bundle work upcoming, it will be a custom Tornado server12:31
gary_postermarcoceppi, btw, one of hatch or Makyo will join you guys in doc sprint when their day starts in a few hours12:32
luca__can someone tell me the commands in the CLI to import and export a bundle or charm to Juju?12:50
luca__gary_poster: rick_h ^12:51
rick_hluca__: I'm not sure I've not tried it out yet :/12:51
rick_hbac: or marcoceppi do you know?12:51
luca__rick_h: useless! :P12:51
gary_posterluca__, you use the juju deployer, not the CLI12:51
rick_hluca__: as usual :)12:51
gary_posterI mean, that is the CLI, not juju itself :-)12:52
marcoceppiluca__: there is no cli yet, though I'm trying to write a plugin for it12:52
marcoceppiFor exporting an environment12:52
marcoceppiimporting you use juju-deployer12:52
gary_postermarcoceppi, I thought Kapil was working on the exporter, or had one?12:52
luca__so if I want to import and export something now via the terminal, how do I do that?12:52
marcoceppigary_poster: maybe? we had jitsu export, but those aren't compatible IIRC12:53
luca__I want the commands for the juju.ubuntu.com website12:53
gary_postermarcoceppi, yeah I thought Kapil was adding it to deployer.  maybe a good idea to sync up with him12:53
marcoceppigary_poster: ack, thanks12:53
gary_posterluca__, there is no export command at this time12:53
luca__gary_poster: right12:53
luca__gary_poster: ok12:53
luca__gary_poster: and is there a way to interrogate the charm store via the command line? like search for stuff? see a list of charms?12:53
marcoceppiluca__: there's a set of tools called `charm-tools` that can search the store and list charms12:54
rick_hluca__: oh oh, this one I do know :P12:54
marcoceppiluca__: they're being rewritten to be executed as `juju charm search`, etc12:54
rick_hmarcoceppi: oh, heads up, api3 with bundle support is coming soon12:54
rick_hmarcoceppi: in case this is what's using the charmworld api12:54
marcoceppirick_h: I have no idea what that means, but is sounds awesome12:54
gary_posterluca__, for import...I think it is nasty: you have to get a certain branch (https://code.launchpad.net/~juju-deployers/juju-deployer/darwin) and then...I forget12:55
gary_posterI mean, nasty in comparison to "install package foo"12:55
luca__marcoceppi: how soon will that happen?12:55
marcoceppiluca__: you can install charm-tools right now and use them as `charm search`, `charm list`, etc12:55
frankbanguihelp: could anyone please review https://codereview.appspot.com/13424043 ? thanks!12:55
rick_hmarcoceppi: I linked you to http://bazaar.launchpad.net/~juju-jitsu/charmworld/trunk/view/head:/docs/api.rst so it means a nwe version of the api with bundles support is coming 12:56
marcoceppiluca__: it's got a lot of tools actually, ppa:juju/pkgs to install charm-tools12:56
rick_hfrankban: looking12:56
frankbanrick_h: thank you12:56
marcoceppirick_h: ah, awesome! will v2 stay around for a while?12:56
rick_hmarcoceppi: yes, but won't get bundles so limited to old scope12:56
marcoceppirick_h: ack, thanks for the heads up12:56
gary_posterluca__, this is pertinent to your deployer import questions: https://docs.google.com/a/canonical.com/document/d/1i_sdbswSovHygc8URsh30zTBx_2G9VZBLlb3qbUmfCM/edit#heading=h.8dx2qatdl27912:57
gary_posterluca__, prerequisites section and #1 of "Steps to be done in front of  a live audience"12:58
luca__gary_poster: I see, thanks13:00
rick_hhatch: morning, want to chat when you get your morning cup of joe please13:03
hatchrick_h: prolly won't be for another 3013:04
rick_hhatch: np, just putting in first dibs since you stuck your head up :P13:04
gary_postermorning hatch. :-)  You and/or Makyo are scheduled to join in doc sprint today.  I'm loathe to lose either of you, but given constraint and inspector work, I'm inclined to ask Makyo to do it.  That postpones his Go relation work.  Is that ok?13:04
abentleybac: I'm documenting the 3.0 API.  It looks like a 3-part bundle-id is always treated as owner/basket/bundle, never basket/revision/bundle.  Do we consider this a bug?13:07
hatchgary_poster: yep that'll be fine, I would like to push to get the inspector unflagged13:07
gary_postergreat hatch, thanks13:08
benjibac: I think you worked on the view that renders /bundles/{basket}/{bundle}/json.  Is that right?13:08
hatchfrankban: thanks a ton for the QA/review I'll look into fixing that shortly13:08
frankbanhatch: np13:09
bacbenji: yes13:09
rick_hbenji: abentley we have a view /json? 13:09
hatchluca__: antdillon hey guys looks like the place in Heathrow that sells sims is in terminal 1 security and I land in 3, do you know of anywhere else close that I should go to pick up a sim card?13:09
benjibac: I can't seem to see a view for /bundles/{basket}/{bundle} (no "json"), do you know if that exists?13:09
abentleyrick_h: I didn't think so.13:10
rick_habentley: yea, that was the 'old' api stuff so same to dupe it over 13:10
antdillonhatch, Near terminal 3 or the office?13:11
abentleyrick_h: "same to dupe it over"?13:11
rick_habentley: /same/shame to pull that convention over13:12
abentleyrick_h: +113:12
bacbenji: no it has not been implemented as it was going to be part of the SEO work13:12
hatchantdillon: well I plan on going from the airport to the science and natural history museum (sunday) so i'd like to get it before then but close to the office would suffice as well and I can pick it up Monday13:12
benjibac: ok, thanks for verifying 13:12
frankbangary_poster: I am going to 1) test the charm-upgrade + set builtin-server story, 2) make fixes if required and then 3) merge juju-gui into charmers. is it ok?13:17
gary_posterfrankban, yes, thank you!13:17
bacabentley: i'm not sure about your question based on the URLs we listed in the google docs "spec".13:18
abentleybac: So, on API 3, there's a way to get a specific revision of an unpromulgated charm: owner/basket/revision/bundle.  For a promulgated charm, basket/revision/bundle won't work.  That seem asymmetrical, though perhaps the intent is that you use owner/basket/revision/bundle for that, too?13:25
antdillonhatch, You should be able to walk into any phone shops in london and pick up a sim card. There are hundreds in london (like starbucks)13:27
bacabentley: that is my recollection of the discussion.  if we have a use case for the api to support unrevisioned access let's list it in the doc.13:29
abentleybac: We already have unrevisioned access.  What we don't have is revisioned access with unspecified owner.13:30
bacabentley: how would that work?  revisioned access for a bundle with no specified owner?  two different people could have a given (basket, bundle) at the same revision.  which do we pick?13:35
abentleybac: We would pick the promulgated one, as we do for two-element bundle ids like "basket/bundle".13:36
hatchantdillon: ohh ok :) any certain network I should try and get on?13:41
hatchI'll likely be using primarily data13:41
antdillonhatch, EE is good, I use them. You might be able to get 4G?13:43
hatchahh cool13:43
hatchI think I am stuck on GSM because of the radios in my phone13:43
hatchthey use a different 4g frequency here I think13:43
antdillonhatch, Then you'll be flighting, 4G is faster then most wifi's13:43
hatchyeah that's the same here13:44
antdillonhatch, Ah ok13:44
hatchI just don't want to spend $0.65/txt, $2/min, $15/MB :)13:44
antdillonOf course, that's nuts!13:45
hatchI think I'll add everyone to my G+ and then use hangouts to communicate once I Have data13:46
hatchassuming others have data....13:46
jcsackettrick_h: can i get you to look at https://codereview.appspot.com/13289046 ?13:53
rick_hjcsackett: sure thing13:53
abentleyrick_h: We were going to have "/download" for bundles so that deployer could use those urls directly.  bac changed "/download" to "/json" for consistency with charms.13:54
gary_posterjcsackett, I didn't see you were back.  how are you?13:55
rick_habentley: not following. What's /download for?13:55
jcsackettgary_poster: i'm alright. tho my typing speed is considerably slower. :-P13:55
gary_posterjcsackett, ack.  glad you are on the mend.  not sure if I want to hear exactly what happened or if I would faint. :-)13:55
abentleyrick_h: $juju deployer -c http://manage.jujucharms.com/bundles/~owner/basket/revision/bundle13:57
rick_habentley: ah, interesting. I didn't realize that deployer was grabbing from charmworld vs the store like charms. 13:57
abentleyrick_h: We don't store bundles in the store at all.13:58
rick_habentley: ok, so ignore me if /json isn't part of matching the deprecated api 13:58
marcoceppiFound a weird bug in the GUI13:58
rick_habentley: hmm, so the deployer won't set the json request headers to send the response on headers vs url change?13:58
rick_hmarcoceppi: shoot man13:58
abentleyrick_h: I still think it should be /download, not /json.13:59
rick_habentley: yea, I'd prefer that it send content-type in the request but oh well13:59
abentleyrick_h: putting it in the URL rather than request headers makes it much easier for a user to control.14:00
rick_habentley: yea, understand. 14:00
abentleyrick_h: Also, we can provide a download link that way.14:01
marcoceppirick_h: yesterday I searched and visisted ~marcoceppi/precise/discourse, which took me to ~marcoceppi/precise/discourse-20. So I went to -HEAD because I wanted to refresh it after I updated the charm in the store and watch the magic. Now when I search the URL it takes me to is -21 but the content is still -20 (as well as -HEAD) it's like there's some severe cache or something14:02
marcoceppirick_h: does https://jujucharms.com/~marcoceppi/precise/discourse-21/ show location as -20 ?14:02
rick_hmarcoceppi: yea, looking14:02
rick_hmarcoceppi: and shows head as r64?14:03
rick_hmarcoceppi: maybe ingest error on changes?14:03
hatchrick_h: ready when you are14:03
rick_hmarcoceppi: run proof?14:03
rick_hhatch: I'm good I think.14:04
marcoceppirick_h: just shows a W about hook not executable14:04
hatchrick_h: ok in guichat14:04
marcoceppirick_h: the change was purely hook based, nothing in metadata changed14:04
marcoceppibut the serach results show the right version (commits and downloads)14:04
marcoceppiand takes me to -2114:04
rick_hmarcoceppi: will have to look into staging to see if there's an import error14:04
marcoceppirick_h: ack, gotchya14:04
rick_hmarcoceppi: I'm all confused, the branch is r64, the rev in the file is 1714:05
rick_hmarcoceppi: so not sure where the 20/21 is even coming from tbh14:05
marcoceppirick_h: that's charm-store version14:05
marcoceppirick_h: which is independent of bzr version and local revision file14:06
* marcoceppi holds back on this14:06
rick_hmarcoceppi: k, will check in a sec.14:06
abentleybac: When a 3-element bundle-id is specified, it looks like we treat the three elements as owner, basket, bundle, but then we return only promulgated bundles.  Is this a bug?14:07
adeuringabentley: could have a look here: https://code.launchpad.net/~adeuring/charmworld/spurious-failures-2/+merge/183183 (a bit longer than the core 5 lines I expected yesterday, but not terribly long either...)14:08
abentleyadeuring: sure.14:09
abentleyadeuring: r=me.  Thanks.14:13
bacperhaps it is abentley.  please file a bug or make a note in the doc if you wouuld.14:15
abentleybac: I've filed bug #121894914:26
_mup_Bug #1218949: Cannot retrieve head revision of an un-promulgated bundle over API 3. <charmworld:Confirmed> <https://launchpad.net/bugs/1218949>14:26
bacthanks abentley14:26
rick_hjcsackett: inboud14:35
rick_hmarcoceppi: cool yea seeing the store sees https://store.juju.ubuntu.com/charm-info?charms=cs:~marcoceppi/precise/discourse 2114:37
rick_hbac: gary_poster there's a config conflict in comingsoon it looks like?14:38
bacrick_h: i'll look14:38
gary_posterthan kyou14:38
rick_hthanks bac 14:38
hatchok now where was I14:40
rick_habentley: have a sec to give me a hand searching this issue for marcoceppi?14:44
abentleyrick_h: sure.14:44
rick_habentley: guichat?14:45
abentleyrick_h: yup.14:45
bacrick_h: try now14:46
gary_posterHey Makyo, I want to reply to rogpeppe's email in which he asked me what the ordering change was that broke the GUI.  IIRC, new relations were being sent before services?  That doesn't sound right unless the deployer was involved.  any quick details I can pass on?14:47
Makyogary_poster, units before services; when we were receiving units after services, the delta triggered adding the unit to a ModelList attached to the service.  If it was before, the service didn't exist.  The solution was sorting the deltas 14:48
gary_postercool thanks Makyo, will pass on14:48
rick_hmarcoceppi: so working on it still, it's fine on staging http://staging.jujucharms.com/api/2/charm/~marcoceppi/precise/discourse-21 and the proof errors listing in production appears to only do reviewed charms so filed #1218966, 14:51
_mup_Bug #1218966: proof errors appear to only display for revewied charms <charmworld:Triaged> <https://launchpad.net/bugs/1218966>14:51
rick_hmarcoceppi: have to go to IS now and have them help get at the logs on production to see what error is thrown. Note that, on production, it stopped at r60, so r61 probably introduced something it didn't like. https://manage.jujucharms.com/api/2/charm/~marcoceppi/precise/discourse-2114:52
marcoceppirick_h: thanks, I wanted to blog/share that link eventually, so I appreciate you looking in to it14:52
frankbangary_poster: charmers' charm is up to date14:52
gary_posterawesome, frankban thanks.  Do you want to send out an announcement/request for testing, or do you want me to?14:53
* marcoceppi looks at rev 6014:53
* marcoceppi 61*14:53
frankbangary_poster: please do it, thanks! for the jujucharms.com upgrade i'd wait till Monday14:54
gary_posterack frankban, completely agree :-)14:54
frankbangary_poster: the upgrade being: 1) upgrade-charm and 2) set builtin-server=true14:54
rick_hfrankban: gary_poster does the constraints issue that hatch is working on effect release though? e.g. will users end up not beign able to deploy?14:55
gary_posterrick_h, no.  frankban made a charm release not a gui release14:55
hatchwith the new inspector14:55
rick_hgary_poster: k, ignore me then. 14:55
gary_poster:-) np14:55
rick_hmarcoceppi: heads up that I do see these errors http://paste.mitechie.com/show/1012/ not sure if those are outdated or what. 14:56
hatchmy laptop has been acting up today.......plz don't break plz don't break plz don't break14:56
marcoceppirick_h: redmine and charm-helper are dead and gone, not sure why it's pulling those14:56
frankbangary_poster: FWIW, we can know from the outside if a GUI is run by the Tornado server visiting the /gui-server-info URL (without the trailing slash)14:57
gary_posterah cool frankban good to know, ty14:57
jcsackettrick_h: thanks for the review. good catch on the needed cleanup.14:59
* benji wonders if the call time has changed.15:01
hatchjujugui - guichat 4 mins ago!15:04
Makyohatch, calendar says 9:30?15:05
hatchfor real???15:05
gary_posterjujugui postponed 30 minutes15:05
hatchoops we all missed that15:05
abentleybac or benji: could you please review https://code.launchpad.net/~abentley/charmworld/api3-docs/+merge/183206 ?15:11
benjiabentley: I'll take a look15:12
abentleybenji: Thanks.15:13
Makyojujugui call in 10 kanban now.15:20
rick_habentley: why is bundles/bundleid plural?15:22
abentleyrick_h: I don't know.  I didn't implement it.15:22
rick_habentley: ok15:22
rick_habentley: r=me15:23
abentleyrick_h: Thanks, but benji was looking at that.15:25
rick_habentley: ok15:25
rick_habentley: well, you can have two, figured I'd peek at since I'll be looking shortly15:25
abentleyrick_h: Cool.15:25
benjiabentley / rick_h: well if we have one, that is enough15:25
benjiI did enjoy the ASCII art table, though.15:26
rick_hbenji: abentley bac is there any opposition to dropping the s on bundles/bundleid for consistency ?15:26
abentleybenji: Yay for ASCII art in ReST.15:26
abentleyrick_h: I am in favour of dropping the s.15:26
benjiabentley: if you /really/ like reST tables, have I got something for you: http://pythonhosted.org/manuel/table-example.html#fit-table-example15:28
benji(and that document which shows how to write tests to test documents, itself includes tests to show that the document is correct)15:30
gary_posterjujugui call now15:30
bacrick_h: i am not opposed.15:30
rick_hbac: yay!15:31
gary_posterjcsackett, ping for call if you can come15:31
rick_hbac: can you drive by that on your current work or should I create a seperate branch?15:46
bacrick_h: i'll do it as a follow-on.  i'll make a card15:49
rick_habentley: filed a bug based ont he log from prod. https://bugs.launchpad.net/charmworld/+bug/121900116:32
_mup_Bug #1219001: production error on several charms "KeyError: 'branch_deleted'" <charmworld:Confirmed> <https://launchpad.net/bugs/1219001>16:32
rick_habentley: looks like a bunch of charms are hitting this keyerror and don't get ingested16:32
hatchjujugui ok well if anyone wants to mee up sunday during the day or for supper/after pm me your cell #16:32
abentleyrick_h: saw.  Huh.16:32
gary_posterhatch, I'll probably email you in that case.  won't have data16:33
gary_posterif you miss it np16:33
hatchgary_poster: sure thing - my plan is to head from the airport to get a sim then to the science/history museum16:34
gary_posterok cool16:34
hatchtexts cost me $0.65 so I want to get that new sim asap :)16:34
frankbangary_poster: I've found a charm bug in the wss url handling when a real pyjuju env is used. Will self review, land, and include in charmers ASAP: https://codereview.appspot.com/1343204316:35
abentleyrick_h: I don't understand how we get that far without setting branch_deleted.16:35
gary_posterthank you frankban 16:35
hatchfrankban: so in moving the constraints stuff into the env/store/python.js deploy method that fixes rapi and go sandbox but breaks python sandbox :/16:35
hatchso not sure what's up with that....sandbox must be broken16:36
=== TheRealMue is now known as TheMue
frankbanhatch: ISTM that the python sandbox still expects to be able to call deploy passing an array of constraints16:36
frankbanhatch: maybe?16:36
abentleyrick_h: That data should already be set when charms are enqueued.16:36
rick_habentley: k, did you want the log file?16:37
hatchfrankban: so I put the formatting in the deploy method in python.js so it's breaking somewhere after the rpc call16:37
abentleyrick_h: Yes, please.16:37
rick_habentley: it's 30M compressed so didn't send/attachj it16:37
hatchso must be fakebackend.js is broken then?16:38
rick_habentley: uploading, will take a sec16:38
frankbanhatch: and sandbox receives the rpc call right?16:40
hatchfrankban: I can never remember the sequence of events, putting in breakpoints now heh16:44
rick_hmarcoceppi: got the logs off production there's a bug for it to follow #1219001. Sorry for the issue16:44
_mup_Bug #1219001: production error on several charms "KeyError: 'branch_deleted'" <charmworld:Confirmed> <https://launchpad.net/bugs/1219001>16:44
hatchfrankban: found the problem, it's in the Service model16:54
frankbangary_poster: charmers updated16:54
frankbanhatch: interesting16:54
abentleyrick_h: I don't see any KeyError on branch_deleted since June 13.  I don't think it can be the cause of this issue.16:56
rick_habentley: oh, crap I went to the end of the file and searched backwards for discourse and didn't read the timestamp :/16:57
* Makyo runs to go get the phone unlocked for real.16:58
rick_habentley: oh...then my bad, let me invalid that bug and go back. 16:58
frankbanhatch: constraintsStr getter?16:58
marcoceppirick_h: oh, branch_deleted ?16:58
rick_hmarcoceppi: ignore me, I found an error in the log from a month ago..back to researching16:59
hatchfrankban: yeah - I don't see any other place to modify the constraints data17:00
rick_hwhoa, just lost power...on laptop battery atm. 17:01
frankbanhatch: good luck, and have a nice weekend!17:01
rick_hnote to self, put desktop on UPS17:01
hatchfrankban: could change it back in the fakebackend17:01
hatchfrankban: you too! see you Monday/Tuesday17:01
rick_hyay power is back17:08
hatchI don't know anyone else with an off brand phone with a micro sim hah, guess I'll just have to buy the unlock code and hope it works once I get there :)17:08
rick_habentley: so I don't see anything in the log. I do notice that discourse r64 is listed in the http://manage.jujucharms.com/recently-changed view17:12
rick_habentley: so now wondering if there's a mongo vs ES issue? is recently changed pulled straight from mongo while the charm data is pulled from ES? or not true?17:12
hatchoh sweet check out the new relation lines on the GUI http://www.ubuntu.com/cloud (scroll down)17:24
gary_posterheh cool17:37
* hatch wonders who's repo that's in lol17:38
hatchI actually think they look pretty cool17:38
hatchof course they are missing the labels but the green adds some nice color17:39
hatchit could even indicate relation status (if there is such a thing)17:39
hatchrick_h: so documentation to get the lxc and go stuff up and running?17:40
hatchunless you just want to QA this branch for me and we can do that later17:41
hatch(would be my preference :) )17:41
rick_hhatch: :P I can QA17:48
rick_hhatch: http://marcoceppi.com/2013/07/compiling-juju-and-the-local-provider/ is what I went with17:48
rick_hwell, some of that, not all needed17:48
rick_henvironments is a 2-liner change and juju switch ftw17:48
hatchrick_h: ok just linting and making sure the tests pass then I'll repropose17:49
rick_hhatch: rgr17:49
hatchI just really want to get this landed instead of potentially fighting getting the system up and running17:49
marcoceppihatch: I recommend just using juju-core from ppa:juju/devel, the local provider afterwards is pretty easy17:49
hatchmarcoceppi: so instead of compiling from source17:50
rick_hhatch: yea, don't compile17:50
rick_hI didn't do that part17:50
hatchok cool - I'll keep notes of what I do then and write another blog post about it17:50
rick_hjust ppa and then walked through the rest for environment/starting up17:50
hatchcan't hurt right? :)17:50
rick_hhatch: +117:50
marcoceppihatch: Yeah, that was me being lazy and killing two birds with one blog post17:50
hatchlazy lazy17:51
hatchhey are you still in Japan?17:51
hatchI just realized it's probably 2am there if you are :D17:52
rick_habentley: guess not, both _find_charm and series_charm_changes are on mongo17:53
hatchrick_h: https://codereview.appspot.com/13252045/  this fixes the bug but not frankbanks request to normalize the config formatting17:56
hatchthat's coming soon, just wanted to get that bug qa'd first17:56
rick_hhatch: ok, so I'm doing the -core qa section?17:57
hatchyes plz17:57
hatchrick_h: wouldn't hurt to do the others too, it's pretty quick17:57
rick_hhatch: ok, starting up17:57
jcsackettrick_h: did you QA my branch, or do i need to hunt down QA?18:05
rick_hjcsackett: sorry, did not QA. I can shortly18:06
abentleyrick_h: It looks like something is wrong with the way we find the head revision in mongo.  I don't know whether it's a bug in the data or a bug in the code.  Elasticsearch does recognize that 21 is the head revision.18:12
abentleyrick_h: I thought recently-changed was all-mongo.18:12
=== schwuk is now known as schwuk_away
rick_habentley: yea, I think it is. Both seem all mongo. 18:12
rick_habentley: sorry, back/forth in qa'ing/etc. 18:13
rick_habentley: so the api uses sort=[('store_data.revision', pymongo.DESCENDING)])18:13
rick_habentley: and the charm_changes don't specify a sort :/18:13
abentleyrick_h: This uses elasticsearch: https://manage.jujucharms.com/api/3/search?series=precise&owner=marcoceppi&name=discourse18:14
jcsackettrick_h: cool, thanks.18:14
rick_habentley: right, but https://manage.jujucharms.com/api/2/charm/~marcoceppi/precise/discourse-21 is what I was looking at pulling old data18:14
rick_habentley: that's using charm/_find_charm right?18:14
rick_habentley: we were looking at why https://manage.jujucharms.com/api/2/charm/~marcoceppi/precise/discourse-HEAD returns rev 60 in the changelog and links to disource-2018:15
abentleyrick_h: Right, it's using _find_charm.18:16
marcoceppiSo, when does this UI show up? http://www.ubuntu.com/cloud18:18
abentleyrick_h: And I'm saying I don't know whether it's a bug in the code or the data, but Elasticsearch looks right.  So it's less likely to be a bug in the data.18:19
rick_habentley: right, gotcha18:19
abentleyrick_h: Because we update them in sync.18:19
rick_habentley: I figured since the recent changed had the data it was maybe how the query/data was looked at18:19
abentleyrick_h: Haven't had a chance to look at that implementation yet....18:20
rick_hhatch: I can't deploy anything. I get the error "Could not deploy the requested service." with/without constraints18:20
hatchwhat's the error in the console?18:20
rick_habentley: can I hand this off to you then? I've got to qa hatch and jcsackett's stuff atm and you're closer to the action18:20
rick_hhatch: none :/18:20
rick_hhatch: sure18:21
abentleyrick_h: okay.18:21
rick_hhatch: if it'll load for me18:21
rick_hhatch: invite me in please18:22
hatchok sec18:23
rick_hjujuhelp how would I get debug js files in a charm deployed jujugui?18:47
jcastrohey rick_h 18:47
jcastrodo you have sinzui-like powers for the store?18:47
gary_posterrick_h, look in charm options.  looking for you...18:48
jcastroactually, I mean abentley 18:48
rick_hjcastro: how so?18:48
rick_hgary_poster: yea, don't see any 'debug' charm options18:48
jcastroso we renamed the rack charm to "rails"18:48
jcastrobut this is not showing the proper charm18:48
jcastromims said something like "they need to flush"18:48
gary_posterrick_h, :-( you are right. I think t must be conflated.looking18:49
abentleyjcastro: I don't think he has special powers.  I think he has to hand-craft urls, but I can do that, too.18:49
rick_hgary_poster: i've tried logging into the service and running make devel, but then it can't talk to the back end18:49
abentleyjcastro: You know that the store doesn't have a concept of rename, right?18:49
rick_hgary_poster: I tried to set the back end to go, no sandbox, and I just get ws errors18:50
jcastroabentley: hah, nice.18:50
rick_hgary_poster: I assume I'm missing whatever listens on the ws end for go18:50
m_3abentley: hi18:51
m_3abentley: so we just need to get any cache related to lp:charms/rails and lp:charms/rack regenerated18:51
gary_posterrick_h, I'm guessing source mapping is insufficient18:51
abentleym_3: Why do you say that?18:51
m_3abentley: so we removed lp:charms/rack18:52
gary_posterrick_h, staging turns on debug mode :-(18:52
m_3abentley: and pushed an --overwrite to lp:charms/rails18:52
gary_posterwe should have a separate flagfor that18:52
rick_hgary_poster: yea, ok cool18:52
m_3abentley: the overwrite was from a lp:~charmers/charms/precise/rack/trunk branch18:53
m_3abentley: and there is no longer a ~charmers owned "rack" branch18:53
m_3so flushing those from the cache and re-injesting lp:charms/rails should handle the changes we need18:53
abentleym_3: There is nothing that would make the "rack" charm disappear.  The charm store deliberately never deletes anything, regardless of whether you delete the branch.18:54
m_3abentley: ok, that's secondary to getting it to recognize the overwrite of lp:charms/rails18:54
m_3abentley: note that the previous version of lp:charms/rails had bzr revno of like 44... the new one has like bzr revno of 11 or 1218:56
abentleym_3: Are you talking about the charm store or the charm browser?18:57
m_3abentley: both ideally... I haven't tested if the store is seeing the latest... only the browser18:58
* m_3 testing store now18:58
rick_hgary_poster: ping, can you join guichat?19:05
gary_posterrick_h, yes, 60 sec19:06
=== schwuk_away is now known as schwuk
abentleym_3: This looks related to the issue marcoceppi was having; If you search, you find precise/rails-1, but when you try to look up precise/rails-1, you get precise/rails-0.19:14
abentleym_3: I think it is unlikely that clearing squid is going to change that.19:15
m_3abentley: bummer19:17
m_3abentley: any suggestions?19:17
abentleym_3: I am already trying to diagnose this bug.  Your case is helpful, because I can reproduce it on staging, where it should be much easier to diagnose.  But it's on me.  There's not much you can do directly to solve it.19:19
m_3abentley: yeah, marco described his issue... sounds like the same19:23
m_3abentley: thanks!19:23
hatchgary_poster: U. Da. Man.19:32
gary_posterhatch, lol, yay!19:32
abentleyrick_h: I believe you were looking at the API3 version of _find_charm.  The API2 version lacks the sort, and I think that's the code path on production.19:33
rick_hgary_poster: so yea, it needs to be an int and it works in the constraints19:34
rick_hgary_poster: thanks19:34
gary_postercool rick_h ! thank you both :-)19:34
* gary_poster continues writing emails :-P19:35
rick_habentley: oh! yea19:38
rick_habentley: should be able to tell by changing the api version in the same request, /me tries it19:38
Makyojujugui: those going to London, if you want to take the Heathrow express instead of Piccadilly (15m vs. 60m): https://gist.github.com/makyo/639350919:39
rick_habentley: hmm, that didn't get it though https://manage.jujucharms.com/api/3/charm/~marcoceppi/precise/discourse-HEAD19:39
gary_postercool Makyo thanks19:40
* rick_h is going to get lost at least twice19:40
MakyoPhone won't unlock, so I won't be reachable.19:42
bcsallerMakyo: you'll still be able to use public wifi many places though so it can be a help19:43
Makyobcsaller, Yeah, still bringing it, of course, just no need giving out my number.19:43
abentleyrick_h: I don't know what revision production is on, but I am certain it's before 371, because it accepts any random number for revision in API3: http://manage.jujucharms.com/api/3/charm/precise/rails-519:43
Makyo$50 for those stupid unlock codes :P19:43
rick_habentley: ah, nvm then19:44
hatchMakyo: $25 http://www.cellunlocker.net/19:44
hatchMakyo: also thanks for the tip about the express19:44
abentleyrick_h: staging is on r373, so on staging, API 3 pukes if you give it a non-existent revision.19:44
Makyohatch, I have the codes, I mean, I just can't unlock a rooted phone, and I can't unroot it right now.19:44
hatchwhat phone?19:46
hatchyou should be able to unlock it without a code if it's rooted19:46
bacjujugui: at the top of the hour i'm going to turn off the cronjob that pulls new releases to comingsoon in preparation for a demo on monday.19:47
benjisounds good19:47
hatchbac: with the new or old inspector?19:47
bachatch: as it currently is on comingsoon19:48
hatchahh ok that should be fine then19:48
hatchjust can't deploy using rapi is all19:48
hatchwith the new inspector that is19:48
gary_posterhatch, the reason behind this is that the inspector will be demoed Monday in an official capacity (see peeps list)19:49
gary_posterjujugui we don't have times on the spreadsheet.  I'm arriving Sunday morning at 7AM.  Anyone else similar?19:50
hatchgary_poster: if you click on them the time is in the top bar19:51
hatchsomeone formatted it incorrectly19:51
bacgary_poster: you get the direct on AA?19:51
Makyogary_poster, I formatted it to datetime19:51
gary_posterbac, yeah19:51
hatchMakyo: thanks19:51
gary_posteroh thanks Makyo awesome19:51
gary_posterJeff I might not wait for you at the airport, but if you want to hang out we can19:52
gary_posterhatch I mean :-)19:52
hatchI wouldn't wait for me either :P19:52
hatch_rick_h, https://gist.github.com/hatched/e72f3db21afec05592b3 solution :D19:53
jcsackettrick_h: any luck QAing?19:55
hatchbcsaller: Makyo mind taking a peek at that gist I posted above... I need to loop through a list of constraints (of any kind) and convert the numbers to integers but leave the strings alone...this technique works by running parseInt and then comparing the two values with type cooersion - thoughts?19:56
bcsaller checking19:57
Makyohatch, sure19:57
Makyohatch, When I ran `juju set-constraint mem=2G` with core on the command line, the gui reported it as 2048.  This would set it to 2 if I use 2G, correct?19:59
Makyo'...if I use 2G in the gui, correct?'19:59
gary_posterMakyo that might be a commandline trick unsupported by the API19:59
rick_hjcsackett: bah, sorry. hatch's qa'ing ran crazy and I didn't get to it19:59
gary_postereither way we should document it, and maybe make it nicr20:00
hatchMakyoL if you typed 2G in the input then it would send 2G to the back end and fail20:00
hatchwe will need client side formatting for these constraints for sure20:00
Makyohatch, constraints[cons] = parsed would set it to (int)2, right?20:00
gary_posterbut making it nicer might be a separate step for later.  for now "send an int in M" would be nice help text20:00
hatchyes but consVal != parsed in that case20:00
hatchso it would set it to the string value20:00
gary_posterwell, not that exactly :-P20:00
rick_hjcsackett: link me please? pulling back up my lxc20:01
bcsallerhatch: I don't think this is proper, you might want to check if it can parse as a number and apply a mapping but if things like 2G need support then its either fancier or a passthrough 20:01
MakyoOh, sorry, was mistaken about the ==.  Alright.20:01
BradCrittendenjujugui: comingsoon locked down at r99220:01
=== BradCrittenden is now known as bac
hatchbcsaller: the Go backend requires they be ints20:01
gary_posterthank you bac20:01
bacuntil monday this time20:01
hatchbut we also need to support things like 'i386'20:01
hatchso we can't simply parseInt all fields20:01
hatchsome need to be strings, some don't20:01
bcsalleralso I'd do the loop with Object.key(constraints).forEach(...), for..in sometimes ends up with issues20:02
bcsallerthe linter will ask for that ifOwnProp check20:02
hatchnot if i turn that off :P20:02
hatchbut anywho....the technique though....that should work?20:03
rick_hjcsackett: nvm, got it, qa'ing now20:03
hatchI haven't found any situations where it fails20:03
hatchjs's type cooersion works pretty well here20:03
hatchbcsaller: (ps yes I will change it to keys() ) :)20:04
hatchgoing to grab some lunch, thanks for looking at that20:05
rick_hjcsackett: fullscreen doesn't work for me :/20:06
abentleyrick_h, marcoceppi, m_3: I believe I understand what's going on, and I've filed the following bugs: #1219064 #1219061 #121906220:06
_mup_Bug #1219064: API 2 does not ensure the latest revision is selected <charmworld:Triaged> <https://launchpad.net/bugs/1219064>20:07
_mup_Bug #1219061: charm pages do not display the latest revision <charmworld:Triaged> <https://launchpad.net/bugs/1219061>20:07
_mup_Bug #1219062: Old JSON API does not ensure it retrieves the latest revision <charmworld:Triaged> <https://launchpad.net/bugs/1219062>20:07
rick_habentley: awesome...kinda. Thanks for taking that through!20:07
marcoceppiabentley: thanks for looking in to it20:07
abentleybasically, it's the same kind of mistake in three places.20:07
rick_hjcsackett: qa-no-ok let me know if you can't repeat the same issues20:08
gary_posterabentley, all three are either solved in API 3, or will be solved when the GUI is forced to use API 3?20:08
jcsackettrick_h: huh, fullscreen worked earlier, but isn't now. looking.20:09
abentleygary_poster: as of r371, API3 is not affected, but production is not on r371 yet.20:09
abentleygary_poster: The three places are API 2, the old JSON api and the charm browser views, so API3 doesn't directly affect them.20:11
gary_posterack abentley.  I just was verifying that we can effectively solve or invalidate all three issues by converting the GUI to use API 320:12
gary_posterand my takeaway is "yes"20:12
gary_posterso, assuming you agree, thank you :-)20:12
abentleygary_poster: That will fix the GUI, but not charmworld.20:12
gary_posterabentley, only API 2 though, right?  And once the GUI uses API 3 we can regard API 2 as kaput, and related bugs as invalid?20:13
gary_posteror are other important clients using API 2?20:13
abentleygary_poster: I'm not talking API2, I'm talking about the web site  that charmworld provides.20:14
gary_postersorry for misunderstanding abentley.  got it.  thanks20:14
abentleygary_poster: Anyhow, I consider these bugs critical, and trivial, so I expect to fix API2.20:15
gary_posterabentley, oh ok cool.  thanks20:15
* benji falls into a lint pile like quicksand.20:34
* benji pulls himself out via a conveniently placed vine.20:45
benjidoes anyone actually use lbox with charmworld?  the .lbox.check fails with ImportError: cannot import name MAXREPEAT for me20:59
abentleybenji: Sounds like you're trying to use it with the wrong python.  But no, AFAIK, no one uses lbox with charmworld.21:04
benjiI just figured it out as you said that.  I was using a python from my host instead of from my lxc instance.21:05
=== schwuk is now known as schwuk_away
gary_posterOK, I'm outta here.  Talk to everyone Monday, and see some of you!21:31
hatchhave a good one gary_poster - I can email/msg/hangout you when I end up with data21:43
hatchsee ya on the other side21:44
hatchMakyo: still round? did you look into unlocking your phone while it's rooted?21:49
Makyohatch, yeah, instructions are 'unroot, unlock, reroot'.21:49
MakyoSo I'll just stick with wifi.21:50
MakyoBoth the hotel and (I'm assuming) the office will have it.21:51
MakyoI can turn on int'l roaming in an emergency, of course.21:51
hatchyeah would have to be one heck of an emergency for $15/MB lol21:51
MakyoHah, I was thinking for voice, granted :)21:52
MakyoA True Emergency21:52
MakyoIn case I need to dial 0118 999 881 999 119 725 3.21:53
baci hope you didn't do that from memory21:54
Makyobac, No, but now it's stuck in my head.21:54
hatchlol what is that number?21:56
MakyoA joke :P  Emergency is 999, that's from IT Crowd.  http://www.youtube.com/watch?v=ab8GtuPdrUQ22:01
hatchohhh :)22:03
* bcsaller places computer in backpack to leave for airport22:28

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