| === eu is now known as Guest92369 | ||
| === Guest92369 is now known as adorilson | ||
| === eu is now known as Guest21731 | ||
| === rodrigo__ is now known as rodrigo_ | ||
| === rodrigo__ is now known as rodrigo_ | ||
| mandel | morning all! | 09:51 |
|---|---|---|
| mandel | dobey, on it! | 09:51 |
| aman_ | how to move ubuntu one default folder to another partition in 11.10 | 09:52 |
| JamesTait | Goooooood moooooooornin' all! | 09:58 |
| amandeep | how to move ubuntu one default folder to another partition in 11.10 | 10:00 |
| Chipaca | amandeep: what is it you're wanting to do? | 10:11 |
| amandeep | i want to move it to a drive with lots of space | 10:12 |
| amandeep | ??? | 10:17 |
| facundobatista | amandeep, currently we only support folders under your home directory | 10:18 |
| amandeep | Thnx alot for the info.... | 10:21 |
| mandel | JamesTait, can you do me a favour and try something in your machine for me? | 10:21 |
| JamesTait | mandel: As long as it doesn't involve rm -rf anything, yes. ;) | 10:22 |
| mandel | JamesTait, I'm having some issues with squid and I wonder if it has to do with my machine setup or something :) | 10:22 |
| mandel | JamesTait, can you please install squid in your system? | 10:22 |
| mandel | JamesTait, Let me pass you a config file I have so that you can try it | 10:23 |
| JamesTait | mandel: I actually have squid installed on my other machine - that's a Lucid box though, is that any good? | 10:23 |
| mandel | JamesTait, sure, I just want to find out why the bloody things does not start when you pass -f | 10:23 |
| mandel | JamesTait, let me paste the config file I'm using | 10:23 |
| mandel | one sec | 10:23 |
| mandel | JamesTait, can you please use the following config http://paste.ubuntu.com/745758/ | 10:26 |
| JamesTait | mandel: I might have to create a couple of directories, bear with me. | 10:31 |
| mandel | JamesTait, yes, the temp/spool/squid and the temp/squi | 10:33 |
| mandel | temp/squid sorry | 10:33 |
| mandel | JamesTait, once you have those two you can do squid -z -f config_file | 10:33 |
| JamesTait | Exactly. :) | 10:33 |
| JamesTait | mandel: I get just a single line of output: 2011/11/22 10:34:24| Creating Swap Directories | 10:35 |
| JamesTait | mandel: Let me see if there's a useful log somewhere. | 10:35 |
| mandel | JamesTait, that is correct, it creates the cache dirs, after than try | 10:36 |
| mandel | squid -N -f config_file | 10:36 |
| mandel | JamesTait, actually add -X to that | 10:36 |
| mandel | so squid -N -f config_file -X | 10:36 |
| mandel | JamesTait, does it work? cause I have a stupid rename error :( | 10:36 |
| JamesTait | Aborted,(core dumped) | 10:36 |
| JamesTait | Wit lots of "2011/11/22 10:36:39| storeUfsDirGetNextFile: opendir: temp/spool/squid/0F/FF: (2) No such file or directory" | 10:37 |
| JamesTait | mandel: Just checking scrollback to see if there's anything more useful than that. | 10:38 |
| mandel | JamesTait, that is not a surprise since you have no cache (although the stupid file is present) | 10:39 |
| mandel | JamesTait, I get FATAL: storeUfsDirCloseTmpSwapLog: rename failed | 10:39 |
| mandel | which I do not understand, and if I run squid in the default location it works.. | 10:39 |
| JamesTait | Permissions? | 10:40 |
| JamesTait | "FATAL: storeUfsDirCloseTmpSwapLog: rename failed" you say? | 10:41 |
| JamesTait | I get that right after all those other entries. | 10:41 |
| mandel | JamesTait, so you ahve the same issue.. WTF | 10:43 |
| mandel | I hate squid | 10:43 |
| * mandel is not a sysadmin for a reason | 10:43 | |
| JamesTait | mandel: http://paste.ubuntu.com/745780/ | 10:46 |
| mandel | JamesTait, yeah, I have the same.. let me look more into this.. | 10:48 |
| mandel | puto squid | 10:48 |
| JamesTait | Maybe get strace on the case. | 10:49 |
| JamesTait | Well that didn't help. rename("temp/spool/squid/swap.state.new", "temp/spool/squid/swap.state") = -1 ENOENT (No such file or directory) | 10:51 |
| JamesTait | Maybe it need absolute paths. | 10:51 |
| JamesTait | mandel: Absolute paths seems to have worked. | 10:52 |
| Chipaca | facundobatista: is it me, or did amandeep go off in a huff? | 10:56 |
| facundobatista | Chipaca, we are clearly not solving his/her problem | 10:56 |
| gatox | hi | 11:01 |
| facundobatista | Hola gatox | 11:04 |
| gatox | facundobatista, how are you this morning? | 11:06 |
| facundobatista | gatox, fine, you? | 11:06 |
| mandel | JamesTait, really? WTF | 11:06 |
| mandel | JamesTait, la madre que los pario.. well, I own you a big one! | 11:07 |
| gatox | facundobatista, fine..... trying to wake up :P | 11:07 |
| mandel | gatox, morning! | 11:07 |
| gatox | mandel, hello! | 11:07 |
| mandel | gatox, did one of your reviews, I have to look at the next one | 11:08 |
| gatox | mandel, great! thanks!....... trying to fix the bug yesterday i felt like coding an onion...... so many layersssss :P | 11:08 |
| mandel | gatox, we really really need to simplify that | 11:10 |
| JamesTait | mandel: You're most welcome. :) | 11:10 |
| gatox | mandel, i agree | 11:10 |
| mandel | JamesTait, well, now I know how to start the bloody proxy service then :) | 11:12 |
| JamesTait | mandel: My main gripe with squid-deb-proxy has been that squid isn't IPv6-capable - at least not the version in 10.04. | 11:15 |
| mandel | JamesTait, squid is just one of the versions of proxies we have to tests.. II guess I'll write a test case for an ISA :( | 11:16 |
| mandel | it hurst just to think about it :P | 11:16 |
| Chipaca | it could be worse | 11:16 |
| Chipaca | it could be an ISA *BUS* | 11:16 |
| JamesTait | Hey! I still have a firewall box with an ISA bus! | 11:17 |
| Chipaca | JamesTait: right! you're in charge of writing the proxy tests | 11:17 |
| JamesTait | Anyway... tea required. | 11:17 |
| Chipaca | mandel: j/k. it's still you, all the time. | 11:18 |
| JamesTait | Actually, if you're going to be like that, maybe something stronger. | 11:18 |
| JamesTait | :-P | 11:18 |
| mandel | haha | 11:18 |
| mandel | Chipaca, don't play with my feelings | 11:18 |
| * JamesTait needs to hit up some Windows developers. | 11:19 | |
| JamesTait | Anyway. TEA! | 11:19 |
| JamesTait | I mean it this time! | 11:19 |
| Chipaca | mandel: ok :) | 11:19 |
| Chipaca | mandel: how're the ribs? | 11:19 |
| mandel | Chipaca, fine, I'm back to 100% :) | 11:20 |
| Chipaca | mandel: w00t :) | 11:20 |
| mandel | Chipaca, I'm trying to look at getting a empty port to be used by the squid proxy.. I'm thinking of using socket.bind(('', 0)).getsockname()[1] is there a better way? | 11:48 |
| mandel | Chipaca, I know there can be a race condition between getting the port and starting squid.. also if you pass 0 to squid it hates you :( | 11:48 |
| Chipaca | mandel: trunk-2a/utilities/utils.py/get_arbitrary_port | 11:49 |
| mandel | Chipaca, which project? | 11:49 |
| Chipaca | mandel: IOW, yes, except use 'localhost' instead of '' | 11:50 |
| Chipaca | because '' has issues with IPv6 | 11:50 |
| mandel | ok | 11:51 |
| duanedesign | morning all | 12:04 |
| gatox | duanedesign, hi! | 12:05 |
| duanedesign | how are you today gatox | 12:07 |
| gatox | duanedesign, really fine, finishing with a bug, you? | 12:08 |
| duanedesign | gatox: oh cool. Which bug if i may ask? | 12:08 |
| gatox | duanedesign, one about confirmation code when the user cancel the registration process and then try to login using the same email.... the hardest part is already done :D | 12:10 |
| duanedesign | ahhh | 12:10 |
| duanedesign | great! | 12:11 |
| gatox | :D | 12:11 |
| karni | gatox: how do you fix that bug? | 12:17 |
| duanedesign | very carefully ;) | 12:18 |
| gatox | duanedesign, jejeje | 12:18 |
| gatox | karni, i'm talking about the client..... when the user execute the installer and choose to sign in with the email already registered, and the server answer not validated email, i change the actual page to verification code page | 12:19 |
| * duanedesign runs from karni | 12:19 | |
| gatox | karni, and change the flow of the wizard to continues from there | 12:19 |
| karni | gatox: ok. so we have the same bug in the wizard I just wrote for Android ;P thanks. | 12:20 |
| gatox | karni, jjejeje | 12:20 |
| gatox | it seems | 12:20 |
| karni | gatox: so when user attempts to register with same username+password, and the registration returns "email already used", you use that username+password to make a Basic Auth authenitcated call to login.ubuntu.com/.../?ws.op=me just as if the user was already registered? | 12:26 |
| gatox | karni, no..... when the server response that the email is already used but not validated, i show the verification page in the wizard, and that page already has the implementation that collect the user email and password and the code that the user should enter in that page and then try to send a request to validate that email as usual (if the user had completed the registration process previously) | 12:28 |
| karni | gatox: collect username and password from where? you guys store users password? | 12:30 |
| karni | gatox: also, does the ws.op=register response contain information that email is used *but not validated* ? I have only seen SSO return "E-mail already used." | 12:30 |
| gatox | karni, no, in the current user sign in page, the user type her email and password in order to login..... and the verification code page, can ask the wizard the info that the password or email fields has | 12:30 |
| gatox | its the same application wizard..... different pages | 12:31 |
| karni | gatox: ah, that is perfect | 12:31 |
| karni | gatox: so you do use the same login and password, okey | 12:31 |
| gatox | yes | 12:31 |
| gatox | karni, the real problem was to intercept the not validated message and send a message to the installer to show the proper page..... the rest it's just reusing the current logic | 12:32 |
| karni | gatox: which URL does give you a response for "not validated"? is it login.ubuntu.com/api/1.0/something?ws.op=me ? | 12:34 |
| gatox | karni, let me show you the code..... | 12:34 |
| * gatox looking... | 12:34 | |
| karni | gatox: Thanks | 12:35 |
| gatox | karni, look here: http://bazaar.launchpad.net/~ubuntuone-control-tower/ubuntu-sso-client/trunk/view/head:/ubuntu_sso/main/__init__.py#L88 | 12:36 |
| * karni loks | 12:36 | |
| * karni looks ;d | 12:36 | |
| gatox | in the login you have a callback to success that check if the email is validated | 12:36 |
| czajkowski | Chipaca: http://www.h-online.com/open/news/item/Canonical-dropping-CouchDB-from-Ubuntu-One-1382809.html | 12:36 |
| gatox | karni, is that what you were looking for | 12:38 |
| gatox | ? | 12:38 |
| karni | gatox: I'm looking for the code of processor.is_validated(credentials) | 12:38 |
| gatox | karni, do you know which object is that?? give me a sec, i'll tell you | 12:39 |
| karni | gatox: I've no idea, it's line 103 form your link. thanks! | 12:39 |
| gatox | karni, ubuntu_sso.account.Account | 12:41 |
| karni | gatox: thank you! | 12:41 |
| gatox | karni, let me know if you need something else..... i'm still with that issue | 12:42 |
| karni | gatox: yes, it uses sso_service.accounts.me() :) thank you :) | 12:42 |
| gatox | :D | 12:42 |
| === rodrigo__ is now known as rodrigo_ | ||
| ralsina_ | good morning! | 13:00 |
| gatox | ralsina, hi! | 13:02 |
| ralsina_ | mandel: ping | 13:02 |
| mandel | ralsina_, pong | 13:02 |
| ralsina_ | mandel: I see nothing happened with the windows release? Is there a problem? | 13:03 |
| mandel | ralsina_, let me check, I have not look at the rt today | 13:04 |
| ralsina_ | mandel: I have :-) | 13:04 |
| mandel | ralsina_, and we do not have the .exe yet? | 13:05 |
| * mandel looks | 13:05 | |
| ralsina_ | mandel: I uploaded the unsigned exe on friday | 13:05 |
| ralsina_ | mandel: https://rt.admin.canonical.com/Ticket/Display.html?id=49209#txn-783031 | 13:05 |
| ralsina_ | mandel: ok, seems like the 49210 RT is not done because the installer has not been signed and published | 13:06 |
| ralsina_ | mandel: I'll re-take it | 13:06 |
| mandel | ralsina_, yes, that I know I was asking question yesterday about that | 13:07 |
| ralsina_ | mandel: got any answers? | 13:07 |
| mandel | ralsina_, no, that he did not know and that there are the super-losas de ones that do the signing | 13:08 |
| mandel | ralsina_, I don't know why on Friday it was market as incomplete.. | 13:09 |
| mandel | and they changed from GSA to LOSA.. no idea why | 13:09 |
| ralsina_ | it has to change to l-sa when it's time to put it on the web, looks like a miscommunication | 13:10 |
| ralsina_ | it moved too quick and fell on the floor | 13:10 |
| ralsina_ | mandel: fixed, it's top of their queue now | 13:11 |
| mandel | ralsina_, ok cool, ping me if they let you know and I'll open the other rt with the correc txml | 13:14 |
| ralsina_ | mandel: ok | 13:14 |
| Chipaca | czajkowski: yep, saw that | 13:24 |
| Chipaca | czajkowski: thanks | 13:24 |
| czajkowski | Chipaca: wasn't me was my boss, but tis interesting read | 13:25 |
| * mandel hates squid sooooo much | 13:26 | |
| ralsina_ | mandel: fry it! | 13:27 |
| ralsina_ | mandel: I know all about squid, so feel free to ask me | 13:27 |
| mandel | ralsina_, awesome, so why does this subprocess.Popen(['squid', '-f %s ' % os.path.abspath('./squid.conf')] fails telling me it cannot find squid.conf when it is there | 13:29 |
| ralsina_ | mandel: probably it's not being executed where you think it is | 13:29 |
| mandel | ralsina, but is an abspath | 13:30 |
| ralsina_ | mandel: yes, but it's abspath of "./squid.conf" which is the cwd | 13:30 |
| ralsina_ | So if it's executing in the "wrong" place, the abspath is wrong too | 13:30 |
| mandel | ralsina, I'm doing that from the same dir as the .conf from a python RELP | 13:32 |
| ralsina_ | print os.stat(os.path.abspath('./squid.conf')) | 13:32 |
| ralsina_ | mandel: trust the computer :-) | 13:33 |
| mandel | ralsina_, I get the stat result no problems what so ever | 13:33 |
| ralsina_ | Ok, the other problem is that you are using Popen wrong, but I need to check the docs ;-) | 13:34 |
| ralsina_ | mandel: subprocess.Popen(['squid', '-f', os.path.abspath('./squid.conf')]) | 13:34 |
| mandel | ralsina, ah, thx! it was that.. what a day.. | 13:35 |
| mandel | ralsina, we do have an issue then in the dbus runner too | 13:35 |
| ralsina_ | mandel: if it had all been a single string it would have worked too | 13:36 |
| Chipaca | czajkowski: i had no idea you were related :) | 13:36 |
| Chipaca | czajkowski: s/related/involved/ | 13:36 |
| mandel | ralsina_, momento estupido :P | 13:36 |
| ralsina_ | mandel: what's the problem with the dbus runner? | 13:37 |
| Chipaca | czajkowski: it's been picked up all over the twitterverse | 13:39 |
| czajkowski | Chipaca: it's interesting to see future plans are underway | 13:39 |
| mandel | ralsina, code style things, no to worry atm | 13:39 |
| ralsina_ | mandel: ack | 13:39 |
| czajkowski | Chipaca: interesting to see no comment from couchbase | 13:40 |
| Chipaca | czajkowski: is it? | 13:41 |
| Chipaca | czajkowski: (not being cute; seriously don't know) | 13:41 |
| * mandel lunch | 13:42 | |
| czajkowski | Chipaca: see pm | 13:47 |
| * czajkowski steals mandel lunch and runs away | 13:47 | |
| Chipaca | czajkowski: haven't got a pm | 13:47 |
| Chipaca | czajkowski: are you identified? | 13:47 |
| czajkowski | aye | 13:48 |
| alecu | mandel, ping | 14:14 |
| fagan | Hey alecu how are you dude :) | 14:15 |
| alecu | hey fagan! long time no see! | 14:15 |
| fagan | Yeah 3 months been moderately busy | 14:15 |
| alecu | fagan, what have you been working on? | 14:16 |
| fagan | alecu: Well was working to find a job first now working on an ipad game with a few friends | 14:16 |
| fagan | alecu: was at a small company for a while but it didnt really suit | 14:17 |
| alecu | fagan, ipad game! cool | 14:17 |
| fagan | alecu: yeah we have some people from cambridge doing the art for it. | 14:17 |
| gatox | alecu, ping | 14:19 |
| alecu | gatox, pong | 14:19 |
| alecu | gatox, I owe you a review, right? | 14:19 |
| gatox | alecu, how are you this lovely morning? (gatox prepares alecu for a review) | 14:19 |
| gatox | jejejejejee | 14:19 |
| alecu | gatox, no te conviene.... | 14:19 |
| alecu | gatox, I've got 9 branches for review :-) | 14:20 |
| gatox | alecu, wowww | 14:20 |
| fagan | alecu: thats just showing off | 14:21 |
| alecu | fagan, hahahha | 14:21 |
| gatox | alecu, is it my branch for the other day included in that 9? | 14:21 |
| * fagan thinks alecu is splitting things into 1 line branches :D | 14:21 | |
| alecu | gatox, no! I've got 9 branches of mine that need review! | 14:21 |
| gatox | alecu, ahhhhhhhhhh | 14:22 |
| alecu | gatox, they are all the backports of the same bug that touches lots of projects: | 14:22 |
| alecu | gatox, take a look here: https://bugs.launchpad.net/ubuntuone-client/stable-2-0/+bug/692597 | 14:22 |
| ubot4 | Launchpad bug 692597 in ubuntuone-storage-protocol (Ubuntu) (and 22 other projects) "If the user has the date/time/timezone wrong, connection fails (Expired timestamp error) (affects: 18) (dups: 6) (heat: 142)" [Undecided,New] | 14:22 |
| alecu | gatox, and look at "Related branches" | 14:22 |
| * gatox taking a look :P | 14:22 | |
| gatox | alecu, nnaaaaaaaaaaaaaa | 14:23 |
| gatox | alecu, ........ so........ not a lovely morniing for you........ | 14:23 |
| gatox | jejejej | 14:23 |
| alecu | gatox, I've been working on this on and off for the last two weeks... and on friday I was finishing the u1-client branches | 14:24 |
| alecu | gatox, today I've got only the control panel branches left | 14:24 |
| alecu | gatox, but now I need to start chasing reviewers... :-) | 14:24 |
| alecu | gatox, luckily alll the branches are very similar to branches that are already merged into trunk | 14:25 |
| gatox | alecu, well.... let mme know if i can help you with a review | 14:25 |
| alecu | gatox, let's trade! | 14:25 |
| alecu | gatox, so, I still owe you a branch... do you have any other more branches to review? | 14:26 |
| alecu | (besides that one) | 14:26 |
| gatox | alecu, the most important one is the one about unicode...... the other one is really trivial and already has mandel approval.... i prefer that you review this one (that you already have been reviewing): https://code.launchpad.net/~diegosarmentero/ubuntuone-client/unicode-issues | 14:27 |
| alecu | gatox, I'm starting with that right now | 14:28 |
| ralsina_ | alecu, gatox: I can do a coupld oe reviews today | 14:28 |
| ralsina_ | couple of reviews | 14:28 |
| alecu | ralsina_, choose the one you like from this tiny list: https://bugs.launchpad.net/ubuntuone-client/trunk/+bug/692597 | 14:29 |
| ubot4 | Launchpad bug 692597 in ubuntuone-storage-protocol (Ubuntu) (and 22 other projects) "If the user has the date/time/timezone wrong, connection fails (Expired timestamp error) (affects: 18) (dups: 6) (heat: 142)" [Undecided,New] | 14:29 |
| gatox | ralsina, great.... i have a trivial one.... and i'm finishing another one that nessita left me about: trying to log in when an email not validated | 14:29 |
| gatox | ralsina, let me give you the trivial one that it's ready | 14:29 |
| ralsina_ | I have the mgmt call in 45 seconds, but I can do reviews in a couple of hours | 14:29 |
| gatox | ralsina, no problem | 14:29 |
| gatox | ralsina, https://code.launchpad.net/~diegosarmentero/ubuntuone-windows-installer/congratulations-page/+merge/82596 | 14:29 |
| gatox | ralsina, the other one is going to be ready in half an hour.... so i'll give you the link later | 14:30 |
| alecu | gatox, let me know when you finish with that and start reviewing, so I'll guide you on reviewing one of my branches. | 14:35 |
| gatox | alecu, ok..... i'll let you know as soon as i finish with this..... only a couple of tests more, and done | 14:36 |
| alecu | great! | 14:36 |
| * mandel back | 14:43 | |
| mandel | alecu, pong | 14:43 |
| alecu | hi mandel! | 14:44 |
| mandel | alecu, hola! | 14:44 |
| alecu | mandel, two things: an important one, and a not so important one :-) | 14:44 |
| mandel | alecu, how was the holiday, short, right? | 14:44 |
| mandel | alecu, shoot | 14:44 |
| mandel | alecu, but before you say it, I did not know she was your sister :P | 14:45 |
| mandel | hehehe | 14:45 |
| alecu | mandel, fine! I just asked for the day because my wife had a small holiday too. | 14:45 |
| alecu | mandel, hahaha | 14:45 |
| alecu | mandel, the important one: I neeeeeeed you to get me one of the galactica expansions! | 14:45 |
| alecu | mandel, I'll pay it, with blood if necessary. | 14:45 |
| mandel | alecu, ok, email me the one amd I'll find it for you :) | 14:45 |
| alecu | mandel, cool! | 14:45 |
| alecu | mandel, the other issue: how's the "squid integration" coming? had any issues? | 14:46 |
| mandel | alecu, I had a small issue starting squid, mainly me being stupid, so I should have a tests case for today that starts squid and uses it, next would be to extend that testcase so that we can add more ocnfig things | 14:47 |
| mandel | alecu, at the moment starts squid at a random port and that is all, I need to ensure auth config etc | 14:47 |
| alecu | mandel, cool | 14:47 |
| alecu | mandel, perhaps the squid conf could make some urls require an auth and some other urls not require auth | 14:48 |
| alecu | mandel, that way the tests can call a method from the testcase and get the first url or the other | 14:48 |
| mandel | alecu, hm.. good idea, is a matter of finding the correct config for that and how to doit nicely more than anything else | 14:49 |
| mandel | alecu, also ading the tests for the test case :P | 14:49 |
| ralsina_ | mandel: I can do the config for that no problem | 14:49 |
| alecu | mandel, right. In any case, remember that we have people with *lots* of squid experience in canonical | 14:49 |
| alecu | mandel, like ralsina_ :-) | 14:49 |
| ralsina_ | hehe | 14:49 |
| mandel | ralsina_, as soon as I have the test case we can take a look at it, since I'm squid agnostic :) | 14:50 |
| alecu | mandel, or life-less, that was (or is) a squid maintainer. | 14:50 |
| mandel | alecu, I'll keep you updated, once is there we can start writing some tests :) | 14:54 |
| mandel | me | 15:00 |
| mandel | ralsina, alecu, gatox , dobey ? | 15:01 |
| gatox | me | 15:01 |
| ralsina_ | mandel: on call, DONE: was sick. TODO: not be sick. BLOCKED: no | 15:01 |
| dobey | heh | 15:01 |
| gatox | jeje | 15:01 |
| dobey | me | 15:01 |
| alecu | me | 15:01 |
| mandel | lets go then | 15:01 |
| mandel | DONE: Progrees on the proxy test case. Got squid running at a random for for the test case. I need to find a nice api for the auth settings. | 15:01 |
| mandel | TODO: More squid test case, specially make it work on windows. gatox review about unicode. | 15:01 |
| mandel | BLOCKED:no | 15:01 |
| mandel | gatox, please | 15:01 |
| gatox | DONE: | 15:02 |
| gatox | Fixed login with not validated account | 15:02 |
| gatox | TODO: | 15:02 |
| gatox | Test for this branch :P, fix more bugs from my bug queue | 15:02 |
| gatox | BLOCKED: | 15:02 |
| gatox | No | 15:02 |
| gatox | dobey, go | 15:02 |
| dobey | λ DONE: fixed u1client SRU, bug #882732 (SRU proposal), reviewed mandel's devtools branch | 15:02 |
| ubot4 | Launchpad bug 882732 in banshee (Ubuntu Precise) (and 3 other projects) "Race Condition setting active source at startup (affects: 5) (dups: 1) (heat: 40)" [High,In progress] https://launchpad.net/bugs/882732 | 15:02 |
| dobey | λ TODO: move some stuff around, review some code, make some awesome | 15:02 |
| dobey | alecu: vamos | 15:02 |
| dobey | λ BLCK: No. | 15:02 |
| alecu | DONE: backporting u1-client branches for bug #692597; day off | 15:02 |
| alecu | TODO: reviews, 1-1, backport u1cp branches | 15:02 |
| alecu | BLOCKED: no | 15:02 |
| ubot4 | Launchpad bug 692597 in ubuntuone-storage-protocol (Ubuntu) (and 22 other projects) "If the user has the date/time/timezone wrong, connection fails (Expired timestamp error) (affects: 18) (dups: 6) (heat: 142)" [Undecided,New] https://launchpad.net/bugs/692597 | 15:02 |
| mandel | ok, I need to go for 30 min, dog needs to walk | 15:07 |
| ralsina_ | alecu: can you post a status update in bug #692597 please? | 15:29 |
| ubot4 | Launchpad bug 692597 in ubuntuone-storage-protocol (Ubuntu) (and 22 other projects) "If the user has the date/time/timezone wrong, connection fails (Expired timestamp error) (affects: 18) (dups: 6) (heat: 142)" [Undecided,New] https://launchpad.net/bugs/692597 | 15:29 |
| ralsina_ | alecu: post it in the bug, I mean :-) | 15:30 |
| alecu | ralsina_, sure | 15:31 |
| ralsina_ | alecu: thanks | 15:31 |
| alecu | te | 15:34 |
| alecu | sorry | 15:34 |
| * mandel back | 16:03 | |
| * gatox lunch! brb | 16:09 | |
| === gatox is now known as gatox_lunch | ||
| dobey | lunch bbiab | 16:29 |
| === ralsina_ is now known as ralsina | ||
| mandel | alecu, ping | 17:01 |
| === beuno is now known as beuno-lunch | ||
| alecu | mandel, pong | 17:03 |
| alecu | gatox_lunch, in expand_user(), in this line: "result = path.replace('~', xdg_home, 1)" | 17:04 |
| mandel | alecu, on question, how are we getting the proxy settings from the system? | 17:04 |
| alecu | gatox_lunch, what happens if path contains the "~" character twice? | 17:04 |
| mandel | alecu, specially taking into account the two approaches | 17:05 |
| alecu | mandel, we are not getting them *yet*, but the plan is: | 17:05 |
| alecu | * on windows or mac, qt gets the settings | 17:05 |
| alecu | * on ubuntu, we get them from gsettings. | 17:05 |
| alecu | mandel, oh, you mean for the integration tests? | 17:05 |
| mandel | alecu, yes, the SquidTestCase needs to patch the methods used for that | 17:06 |
| === gatox_lunch is now known as gatox | ||
| alecu | mandel, for the integration tests, the tests themselves will choose if they need to use: a) no proxy, b) anon proxy, c) authed proxy. | 17:06 |
| gatox | alecu, only replace the first one | 17:06 |
| alecu | gatox, oh, right! I missed the "1" as the last parameter! | 17:07 |
| alecu | gatox, dumb me! | 17:07 |
| alecu | gatox, sorry :-) | 17:07 |
| gatox | alecu, no problem! | 17:07 |
| mandel | alecu, hm.. then in the setUp of the tests you can say, get_aut_proxy_settings for example, otherwise we don't match very well with the current services implementation in ubuntuone-dev-tools | 17:08 |
| mandel | alecu, the DBustTestCase patches the methods that get the Session which makes sense | 17:09 |
| alecu | right | 17:09 |
| alecu | mandel, but for proxies it looks different to me | 17:09 |
| alecu | mandel, my guess is that we should choose to use proxies or not from each individual test | 17:10 |
| mandel | alecu, we would like to be able to get a test case to be ran in the three diff scenarios, right? | 17:10 |
| mandel | alecu, or just a subset? | 17:10 |
| alecu | mandel, that can be an option too: have a test be repeated in each of the three scenarios | 17:11 |
| mandel | alecu, that is the one I was thinking to be the most common one.. | 17:11 |
| mandel | alecu, why don't you write a test with what you would imaging using, I'll do the same and lets see where we get | 17:11 |
| alecu | mandel, but anyway we want to separate the "discovery of the configuration" from the "using the proxy" into different sets of tests. | 17:11 |
| alecu | mandel, so, these tests you are working on are about "using the proxy" | 17:12 |
| alecu | mandel, and I would not want to mix the "discovery of config" into that. | 17:12 |
| alecu | mandel, yup, it makes sense. I'll try writing some tests in half an hour or so. | 17:13 |
| alecu | mandel, if it's past your EOD, I'll mail it | 17:13 |
| mandel | alecu, it probably is, I'll do the same know.. | 17:13 |
| mandel | s/know/now | 17:13 |
| mandel | ralsina, gatox, I'd like your input too ^ | 17:16 |
| * ralsina reads | 17:17 | |
| mandel | ralsina, gatox, it would be nice to have a decent api for that | 17:17 |
| gatox | mandel, i'm lost.... about what? | 17:17 |
| gatox | mandel, from alecu ping? | 17:17 |
| ralsina | +1 that config discovery is a whole different thing | 17:17 |
| gatox | :P | 17:17 |
| alecu | mandel, I think I got disconnected when you pasted *something* | 17:18 |
| alecu | mandel, because I don't understand what you'd like the input on | 17:18 |
| gatox | mandel, alecu +1 to separate those things.... i would be easier to control which one may be causing problems in some point and understand the code | 17:21 |
| mandel | alecu, on how you would like the tests to be written | 17:22 |
| mandel | alecu, ralsina, gatox for example: http://paste.ubuntu.com/746122/ | 17:22 |
| mandel | not that any of those are good ideas | 17:23 |
| mandel | I'm more interested in how you want to write the tests than the implementation... I can think about that later | 17:24 |
| alecu | mandel, http://paste.ubuntu.com/746137/ | 17:30 |
| alecu | mandel, I think that way is simpler | 17:31 |
| alecu | mandel, the tests would only be on the first class; the second and third class would only have a smallish setUp method | 17:31 |
| mandel | alecu, looks nicer, now, how do they get the proxy seetings, like port number etc.. just set them in a self.proxy_settings? | 17:34 |
| alecu | mandel, and setupAuthProxy and setupAnonProxy could be used directly from other tests methods in other testcases | 17:34 |
| mandel | alecu, hm, if you allow that you will have to be careful with the tearDown and cleanUps | 17:34 |
| alecu | mandel, our webclient library will probably use some function to get the settings. Those two methods (setupAuthProxy and setupAnonProxy) should just "self.patch" that function. | 17:35 |
| mandel | alecu, aha, that is what I was think, but ofcourse we don't know that yet | 17:35 |
| mandel | alecu, I need to think how well those that map with the current design of ubuntuone-dev-tools since services are ALWAYS started by the test runner | 17:37 |
| mandel | dobey, ping | 17:38 |
| alecu | mandel, that's great. We want one instance of squid running all the time. | 17:38 |
| dobey | mandel: hi | 17:38 |
| alecu | mandel, it's the responsibility of the twin functions above to let the webclient know whether to use the squid or not. | 17:39 |
| mandel | dobey, I have a question, at the moment the DBusRunner is always ran by the TestRunner, is there a reason for that, rather than doing it in the setUp of the test case? | 17:39 |
| mandel | dobey, I've been talking with alecu about the API that the SquiTestCase should have and he came with a nice idea: http://paste.ubuntu.com/746137/ | 17:40 |
| dobey | mandel: it wasn't working when i did it in the setUp/tearDown; and spawning killing a service for every test is also quite expensive | 17:40 |
| mandel | mm | 17:41 |
| dobey | mandel: yes; you really do not want to spawn a new squid for every single test | 17:41 |
| mandel | alecu, dobey is right, is going to be very expensive to do that | 17:42 |
| dobey | mandel: i'm not sure what the best solution is for that, but this also goes back to the discussion we had yesterday | 17:42 |
| mandel | dobey, I had uglier ideas like: http://paste.ubuntu.com/746122/ | 17:42 |
| dobey | i don't think we want to be using squid in unit tests | 17:42 |
| alecu | mandel, I know. And I'm not proposing to start/stop the squid on each test. | 17:42 |
| mandel | alecu, ok, so we start the service and those methods only patch the x method that gets the proxy settings, that makes sense | 17:43 |
| alecu | mandel, I'm proposing the webclient library using it or not on each test. | 17:43 |
| alecu | mandel, exactly. | 17:43 |
| mandel | alecu, I misunderstood you | 17:44 |
| alecu | mandel, I agree with dobey that we probably don't want this to be run during unit tests | 17:45 |
| alecu | mandel, so probably these tests should be run separately, with a different script | 17:45 |
| mandel | alecu, that is not a big problem, we can later make u1trial ignore them or not in a nice way | 17:46 |
| dobey | maybe devtools is the wrong place to even have this | 17:47 |
| mandel | dobey, ideally it would be nice to have eveything in a central place for all the projects to be tested with proxy.. so where do you think it should be moved? | 17:48 |
| dobey | mandel: a central place for it to exist makes sense; but i am not sure it should be done via the u1trial services magic. mostly becuase i wouldn't want to suggest other people to use it, and having it there does the exact opposite of that :) | 17:50 |
| dobey | wouldn't want to suggest that it's sane for people to have that in their unit tests, that is | 17:51 |
| mandel | dobey, yeah.. but it is so easy to do it there, that I think we are already doing it ;) | 17:52 |
| mandel | dobey, the team chat is tom, right? | 17:52 |
| dobey | mandel: so maybe in devtools is an ok place to have such a thing, but probably via some other means. perhaps a script or set of tools, for doing integration testing with larger instance system services like squid/apache/mysql/etc | 17:52 |
| gatox | ralsina, when you have a moment..... review please: https://code.launchpad.net/~diegosarmentero/ubuntuone-windows-installer/not-validated-account/+merge/83054 - https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/not-validated-account/+merge/83055 | 17:52 |
| dobey | mandel: thursday; but i won't be on it :) | 17:52 |
| mandel | dobey, dammed.. | 17:53 |
| mandel | dobey, so.. I'll read a little of how to do proper integration tests with python, then with all the info I'll see if I have a better idea of how to integrate with devtools | 17:54 |
| dobey | mandel: actually; probably it belongs in ubuntuone-testing, which is where the tests stuff lives that gets run in jenkins | 17:54 |
| mandel | although, I would call what we do with dbus integration tests :P | 17:54 |
| dobey | mandel: well, what we do with dbus is more self-integration tests. we test that the cllient API talks to the server API correctly; and we own both ends of that | 17:55 |
| dobey | mandel: and the dbus service thing is to ensure we don't screw wiht the live session when running the tests | 17:55 |
| mandel | dobey, I know the reason, is just that it looks like that hehe | 17:57 |
| mandel | dobey, I'm looking at the ubuntuone-testing, it uses selenium to run the tests (I have not look that them, but I suppose they are for the web) | 17:58 |
| mandel | is that project open? we could write the test there.. but looks funny to have the client, sso, control-panel tests for proxy there | 17:58 |
| dobey | mandel: right; and i am not happy about that, either. but squid isn't a service that all of GNOME depends on either :) | 17:58 |
| gatox | lisette, did you have any update about this issue: https://bugs.launchpad.net/ubuntuone-control-panel/+bug/878867 ?? (the one i told you yesterday.... i saw your e-mail but didn't get any update) | 17:58 |
| ubot4 | Launchpad bug 878867 in ubuntuone-control-panel "QT UI Inactive button while (re-) connecting in progress (affects: 2) (dups: 1) (heat: 20)" [Medium,Confirmed] | 17:58 |
| mandel | dobey, true.. | 17:58 |
| dobey | mandel: ubuntuone-testing is for all the tests that will run in jenkins | 17:58 |
| dobey | mandel: they simply started with the selenium/web stuff, as it's easier to do | 17:59 |
| mandel | ack | 17:59 |
| dobey | mandel: testing GTK+ UI inside jenkins is a bit harder :) | 17:59 |
| mandel | dobey, I guess I could do something in that area.. I need to go (rugby training) but I'd really appreciate if you can send me your input to my mail, if you have the time that is | 18:00 |
| === beuno-lunch is now known as beuno | ||
| mandel | I'll bring this conversation up on thursday to see wtf we do | 18:00 |
| dobey | mandel: it was always my understanding that testing against live proxy server would only be done in jenkins (or in real-world use/testing), and that only unit tests would be in the client code necessarily | 18:00 |
| lisette | gatox: no, waiting for input from cparrino and roberta | 18:01 |
| gatox | lisette, ok thanks...... one more question | 18:01 |
| lisette | shoot | 18:01 |
| dobey | alecu: ^^ is that not what we agreed on in the call we had from hallway at UDS? | 18:01 |
| mandel | dobey, ah, but I missed that call :( | 18:01 |
| mandel | was on holidays.. | 18:01 |
| gatox | lisette, do you know if there is some contextual help texts done for the wizard?? it's about this bug: https://bugs.launchpad.net/ubuntuone-windows-installer/+bug/831312 | 18:02 |
| ubot4 | Launchpad bug 831312 in ubuntuone-windows-installer "QT UI: implement contextual help re-enabling the ? button (affects: 1) (heat: 4)" [Low,Triaged] | 18:02 |
| lisette | gatox: not yet, but it is on mine and patricia´s radar | 18:02 |
| gatox | lisette, ok thanks! i'll wait then | 18:02 |
| lisette | gatox: sorry.............! | 18:03 |
| gatox | alecu, i've finish with my issue..... i was looking for some other issue, but i recall that you need reviews | 18:03 |
| gatox | lisette, no problem! :D | 18:03 |
| dobey | mandel: but alecu should have said it to you, when you started talking with him about it i guess :) | 18:03 |
| alecu | dobey, mandel: yes, the saner thing would be to use squid on integration tests. | 18:03 |
| alecu | dobey, mandel: that's exactly what's written on the bug: #884963 | 18:04 |
| alecu | bug #884963 | 18:05 |
| ubot4 | Launchpad bug 884963 in ubuntuone-dev-tools (and 1 other project) "proxy integration tests (affects: 1) (heat: 6)" [Undecided,Triaged] https://launchpad.net/bugs/884963 | 18:05 |
| mandel | alecu, says nothing about jenkins ;) | 18:06 |
| mandel | alecu, but is not a big problem, I sometimes miss working in an office... | 18:06 |
| alecu | mandel, it should say nothing about jenkins! | 18:06 |
| alecu | mandel, we need some script that we can run in our computers | 18:06 |
| alecu | mandel, that script does integration tests using squid | 18:07 |
| gatox | alecu, ping..... do you want me to review some of your branches?? or i should take another of my issues? | 18:07 |
| alecu | mandel, and that script can be run by jenkins *too*. | 18:07 |
| alecu | gatox, sure, reviews, please! | 18:07 |
| gatox | alecu, ok...... shoot | 18:07 |
| alecu | gatox, bug #692597 | 18:07 |
| ubot4 | Launchpad bug 692597 in ubuntuone-storage-protocol (Ubuntu) (and 22 other projects) "If the user has the date/time/timezone wrong, connection fails (Expired timestamp error) (affects: 18) (dups: 6) (heat: 142)" [Undecided,New] https://launchpad.net/bugs/692597 | 18:07 |
| gatox | alecu, ok | 18:08 |
| alecu | gatox, there are 5 branches already merged for that bug. | 18:08 |
| mandel | alecu, ok, I get that, then where the bloody hell should this code go? is not unit tests, which is ok, we simply add a diff script and we are done with it but we have to let u1trial know no to run them | 18:08 |
| alecu | gatox, I'm making backports of those branches, and those are the branches "ready for review" | 18:08 |
| alecu | gatox, the main thing is that I'm running the tests for those branches in a Maverick, Natty and Oneiric VMs | 18:09 |
| gatox | alecu, ok....... i can review them in any ordeer? | 18:09 |
| mandel | alecu, so, I'd use u1trial, start squid per testcase like we do with dbus and simply provide a u1trial --rpoxy flag or something | 18:09 |
| alecu | gatox, and also the branches depend among themselves, but it says so in each merge proposal | 18:09 |
| gatox | alecu, this is going to cost you a lot | 18:09 |
| gatox | jejejjejejee | 18:09 |
| alecu | gatox, perhaps you can start with some of the "storage-protocol" branches | 18:09 |
| gatox | ok | 18:10 |
| alecu | mandel, that's a great idea... | 18:10 |
| alecu | mandel, let's discuss this a bit more | 18:10 |
| mandel | alecu, lets use my last 15 min before rugby training :) | 18:11 |
| alecu | mandel, right now we have a script (run-tests) that calls u1trial, and it sets some modules to test | 18:11 |
| alecu | mandel, (we may discuss this tomorrow, let me know when you have to go) | 18:11 |
| alecu | mandel, well | 18:11 |
| alecu | mandel, the thing is that we are also passing some stuff to u1trial that *should not* be tested. | 18:12 |
| alecu | mandel, so perhaps we should be passing the "integration tests" in that list of *should not* | 18:12 |
| alecu | mandel, and conversely, passing the integration tests when calling u1trial from a different script | 18:12 |
| alecu | dobey, does that make sense? ^ | 18:13 |
| mandel | alecu, then, it would be nice for the sake of extensibility to have an --integration flag that will detect which tests are integration tests and then just do a u1trial --integration | 18:14 |
| mandel | alecu, and each new testcase we add can be added to the ignored test cases when we do not have the flag (rather than proxy which is very specific) | 18:14 |
| dobey | alecu: i don't think u1trial is the right place to have this | 18:14 |
| gatox | alecu, conflicts here: https://code.launchpad.net/~alecu/ubuntuone-storage-protocol/timestamp-autofix-1-4/+merge/81758 | 18:15 |
| gatox | moving to another one | 18:15 |
| alecu | mandel, that's a lovely idea, but till we have that we can fake it | 18:15 |
| alecu | mandel, I do not want to block having the integration tests on having a much smarter u1trial. | 18:15 |
| alecu | dobey, then we need something that works like u1trial but that has a different set of plugins. | 18:16 |
| alecu | dobey, u2trial ? | 18:16 |
| mandel | alecu, adding that to u1trial is simple, I just don't know why dobey says u1trial is not the place, that is what is playing with my brains | 18:16 |
| dobey | alecu: i don't think so. | 18:16 |
| dobey | mandel: because it is an abuse of the python unit testing framework, in order to do integration testing, to have it there | 18:17 |
| mandel | dobey, well, but there are no intergration testing frameworks.. and I'm not writing one :P | 18:18 |
| dobey | alecu, mandel: to me, this seems like something we would test more like how our server tests are done, as they require certain services like squid to be up with specific configs. and probably such tests belong in ubuntuone-testing. i think we need to bring elopio and rmcbride into the discussion | 18:18 |
| dobey | mandel: i'm not suggesting writing one. i think we have most of it written anyway :) | 18:19 |
| * rmcbride looks at the backlog | 18:19 | |
| dobey | alecu: also, why do we need these integration tests only with squid? what about the 5 different MS proxy servers? | 18:20 |
| mandel | dobey, squid is a starting point | 18:20 |
| alecu | dobey, we want to test that our code is at least setting up the network library the right way | 18:21 |
| dobey | mandel: so then after squid, we have MSProxyIIS2004TestCase? :) | 18:22 |
| dobey | alecu: i don't think we need squid to test that do we? | 18:22 |
| alecu | gatox, approved btw. | 18:22 |
| rmcbride | OK, that's a lot of backlog | 18:22 |
| mandel | dobey, hopefully not with that ugly name and we can do something smart | 18:22 |
| gatox | alecu, great!!!! | 18:23 |
| elopio | I'm not sure about placing those tests in u1-testing. Currently, the tests there are high level, and they don't depend on the internals of ubuntu one, just on what's visible to the client. | 18:23 |
| dobey | elopio: right, which is why i think proxy integration tests belong there (ie; configure a proxy, make sure client connects to things through that proxy) | 18:24 |
| elopio | sorry, s/client/user | 18:25 |
| elopio | to me, it seems like it should be separate test suite, brother to the unit test suite and in the same project. | 18:25 |
| mandel | elopio, which project do you mean, ubuntuone-client or ubuntuone-testing? | 18:25 |
| elopio | mandel, ubuntuone-client. On ubuntuone-testing we could add some tests, but using mago or xpresser to do the proxy configuration. | 18:26 |
| mandel | elopio, cool we where thinking about that kinda of | 18:27 |
| elopio | at least, that's the current level of the tests. We could change the scope. | 18:28 |
| dobey | elopio: there are two sets of tests here; unit tests belong in unit tests, of course; but i think the integration tests where we configure a proxy and check that things go through that proxy, belong in ubuntuone-testing, no? | 18:28 |
| mandel | alecu, elopio, dobey, rmcbride I need to go to rugby, please continue with the dicussion, I'll read the back log asap :) | 18:28 |
| rmcbride | yea, our current client focus is on getting mago and/or xpresser automation to do things for us. | 18:28 |
| * mandel goes to break some bones | 18:28 | |
| mandel | hopefully not my bones :P | 18:28 |
| alecu | mandel, watch out for your own bones | 18:28 |
| alecu | ditto | 18:29 |
| rmcbride | dobey: agreed on checking the proxy works, not sure how the configuration of the proxy is done, so I'm not sure about that one | 18:29 |
| rmcbride | if it's something that the user would have to do manually, it's absolutely something that u1-testing would cover | 18:29 |
| dobey | rmcbride: gnome network config UI; so would be via mago | 18:29 |
| dobey | well, the user would do it manually, at some level probably | 18:30 |
| rmcbride | right. that's waht I wasn't getting from the recent discussion | 18:30 |
| rmcbride | I have no idea how the whole | 18:30 |
| dobey | i don't think proxy config in dhcp responses get handled well by things | 18:30 |
| rmcbride | "use a proxy" | 18:30 |
| rmcbride | work flow will fucntion | 18:30 |
| dobey | configure proxy, stuff doesn't break; <- this is how it should function :) | 18:31 |
| elopio | dobey, they can be there. It's not a big problem. But I think that mandel's tests are more like the integration tests that facundo made in the servers project | 18:31 |
| elopio | lib/ubuntuone/storage/tests/integration | 18:31 |
| rmcbride | dobey: I mean how client is interacting. The only thing I'm getting from the discussion is "we should have these tests somewhere and they may or may not involve squid" | 18:31 |
| * rmcbride hasn't been reading this thread for as long as it has been going on | 18:32 | |
| dobey | elopio: i think that's yet to be determined, as we don't actually have any tests yet :) | 18:32 |
| elopio | those are too coupled with the project internals. Are more like unit tests than like UI tests. | 18:32 |
| dobey | rmcbride: well, squid is "we need a proxy server" | 18:32 |
| elopio | dobey, ok, then I think that there should be integration tests at the level of those server tests :) | 18:32 |
| dobey | rmcbride: in CI, we probably want to have several different proxy servers running, and test that the client works correctly with all of them | 18:32 |
| elopio | and then, integration tests at the level of the user. That's u1-testing. | 18:33 |
| rmcbride | dobey: ah yes. | 18:33 |
| dobey | elopio: i don't think these are coupled with the code internals at all | 18:33 |
| alecu | dobey, I think that lower level tests may be coupled with the code | 18:34 |
| dobey | alecu: i don't think lower level tests need an actual proxy server | 18:34 |
| elopio | dobey, if they are not coupled, I see no problem in placing them in a different project. It can be u1-testing, or u1-integration-testing, or anything else. | 18:34 |
| rmcbride | OK so basically, for the CI stuff that we would be covering with u1 testing, we'll need a proxy server in the environment, and eventually proxy use configuration as one story in the various test suites | 18:35 |
| alecu | dobey, not unit tests, but integration tests similar to the u1client-server tests. | 18:35 |
| rmcbride | for platforms | 18:35 |
| rmcbride | bear in mind | 18:35 |
| elopio | I see three levels: unit, integration and user. | 18:35 |
| rmcbride | we don't yet have smoketest done apart from webUI | 18:36 |
| elopio | if you can fake or mock all the proxies, we can test almost everything in the unit level, and the integration level might not be necessary. | 18:36 |
| dobey | alecu: why? | 18:36 |
| rmcbride | elopio: agreed on that. | 18:36 |
| elopio | but it also depends on what we can test at the user level. | 18:37 |
| alecu | dobey, because they catch different type of errors that unit tests don't catch. And are more fine grained than high level integration tests. | 18:37 |
| rmcbride | elopio: yep. That's what I am unclear on from the discussion so far | 18:37 |
| dobey | alecu: then maybe those belong in ubuntuone-servers; because the client branches can't spawn a server to talk to as well | 18:38 |
| dobey | alecu: if they're going to be end-to-end tests, they need to be somewhere that is possible | 18:38 |
| elopio | dobey, that makes sense. If we need two dbus sessions to test the synchronization through firewalls, those tests seem more like brothers of the ones facundo did on the servers project. | 18:39 |
| alecu | dobey, hmm... I don't think a full server is needed. Just an ssl enabled port that answers with the ubuntuone-storage-protocol handshake | 18:40 |
| ralsina | Sorry, was on a call. What should I decide? :-) | 18:41 |
| alecu | dobey, and that's completely doable with twisted on the client tests | 18:41 |
| elopio | ralsina, you should say: "let's do all the kinds of tests, the more the merrier" | 18:41 |
| alecu | and having them on ubuntuone-client will mean that they will get run a lot more often by reviewers. | 18:41 |
| ralsina | what elopio said! | 18:41 |
| * ralsina cracks the whip, for emphasis and great justice | 18:42 | |
| * dobey goes to get some bourbon | 18:42 | |
| dobey | better start drinking now | 18:42 |
| ralsina | dobey: ha | 18:42 |
| rmcbride | dobey: polished off that Cuban Rum already? | 18:42 |
| dobey | rmcbride: finished it in orlando | 18:43 |
| rmcbride | dobey: ah. Shame I wasn't on site longer I guess | 18:43 |
| ralsina | some integration testing is always useful, It's conceptually different from unit testing. Unit testing tests "dos it do what it's supposed to", integration tests are "is it trying to do the right thing?" | 18:43 |
| dobey | well, i understand beuno has a nice stock we can clean out | 18:43 |
| elopio | my opinion (almost ignorant as I have just read a little of the unit tests) is that on ubuntuone-client we should have tests that check that the client can talk with all the proxies, through fakes. | 18:44 |
| elopio | on ubuntuone-servers, or new project ubuntuone-integration-testing, we should have tests that check that two clients can be synchronized through firewalls. If they are the real firewalls, that's better. | 18:44 |
| alecu | elopio, I would agree with that, but since the code of the libraries that do the network access (qt-network and libsoup) is not python, we cannot use mockers there. | 18:45 |
| elopio | and on ubuntuone-testing, we should have tests that check that two different users can synchronize their files through the UI. | 18:45 |
| alecu | elopio, I completely agree with the second point | 18:46 |
| dobey | better would be firing all the IT persons in the world who ever suggest setting up a proxy (in this sense) and requiring users to go through it, as a solution | 18:46 |
| elopio | alecu, sure we can. We can add use our own interface on python to talk to those libraries. | 18:46 |
| elopio | then we fake the responses. | 18:46 |
| dobey | alecu: it's Python. you can do anything in Python. | 18:47 |
| dobey | that's what makes it so great! | 18:47 |
| dobey | </sarcasm> | 18:47 |
| gatox | alecu, i've some really bad news for you..... please sit down..... are you sit? :P | 18:47 |
| alecu | gatox, sure | 18:47 |
| gatox | alecu, all of your branches have conflicts | 18:47 |
| alecu | gatox, ouch! | 18:48 |
| gatox | alecu, i added the message in each one | 18:48 |
| gatox | the comment | 18:48 |
| gatox | i mean | 18:48 |
| alecu | gatox, are you merging them with stable? | 18:48 |
| gatox | yep | 18:48 |
| gatox | i branch trunk locally...... and then merge with yours | 18:48 |
| alecu | elopio, I'm not sure I'm understanding "We can add use our own interface on python to talk to those libraries." | 18:49 |
| dobey | anyway; i don't think configuring/starting a squid server (or any server on that level of service actually), belongs in ubuntuone-dev-tools | 18:49 |
| alecu | gatox, fiuuuuuu, you worried me. | 18:49 |
| gatox | alecu, why?? do i did something wrong? | 18:49 |
| alecu | gatox, you should not branch trunk, you should branch each different "stable" branch. | 18:49 |
| elopio | alecu, sorry, that's my crappy english getting in the middle. | 18:49 |
| alecu | elopio, mine's not a lot better :-) | 18:49 |
| alecu | gatox, say, let's start with one of them | 18:50 |
| elopio | it's the principle of isolation. If we are using a external library that we can't control, it's better to add a class we can control in the middle. | 18:50 |
| gatox | alecu, ohhhhh crap..... i mark each one as need fixing..... | 18:50 |
| alecu | gatox, for instance, this one: We can add use our own interface on python to talk to those libraries. | 18:50 |
| elopio | it will just pass messages from us to the library. If we use that, we can fake the library responses. | 18:50 |
| alecu | doh | 18:50 |
| alecu | gatox, this one: https://code.launchpad.net/~alecu/ubuntuone-storage-protocol/timestamp-autofix-1-4/+merge/81758 | 18:50 |
| alecu | elopio, yes, and those are unit tests. | 18:51 |
| elopio | alecu, yes. | 18:51 |
| gatox | alecu, i should branch this: lp:ubuntuone-storage-protocol/stable-1-4 | 18:51 |
| ralsina | You guys know there already are debugging proxies that do that, right? http://www.fiddler2.com/fiddler2/ | 18:51 |
| gatox | and the mmerge with yours? | 18:51 |
| alecu | gatox, exactly! | 18:51 |
| dobey | elopio: we don't need to do anyting special like that to fake the libraries | 18:51 |
| * gatox hits his head with the desk | 18:51 | |
| gatox | alecu, ook...... reviewing again | 18:51 |
| elopio | on the integration tests, it would be better not to fake anything. | 18:52 |
| elopio | dobey, even better then :) | 18:52 |
| dobey | ralsina: that is basically just another proxy server, yes | 18:52 |
| ralsina | dobey: yes, but it doesn't need a server behind it because it can record traffic and fake it | 18:52 |
| alecu | ralsina: that sounds interesting, but we need protocols other than HTTP(s) | 18:53 |
| dobey | ralsina: yes, but so can squid; so it's baasically the same thing as squid, but instead, it's a windows app that supports less stuff :) | 18:54 |
| elopio | can we configure all the real proxies through bash or python? If that's possible, then we just need to overwrite the setup of facundo's tests to put the two clients behind the proxies. And that's the integration suite. | 18:54 |
| dobey | alecu: do we really need an end-to-end test that only does the protocol handshake here? | 18:55 |
| alecu | ralsina: also, we are very interested in proxy auth, and fiddler does not seem to have any of that. | 18:55 |
| ralsina | alecu: agreed | 18:55 |
| ralsina | elopio: yes, we can | 18:56 |
| ralsina | elopio: at least squid | 18:56 |
| elopio | ralsina, and what do you think? should that be part of ubuntuone-servers? | 18:57 |
| alecu | elopio, ralsina: I think that's a good solution, but we can also do it with tests that run on the client | 18:57 |
| alecu | elopio, ralsina: the problem with the tests that are run on the server side is that our developers seldom run them. | 18:58 |
| elopio | alecu, do you mean at the unit level? | 18:58 |
| ralsina | alecu: that was in part because they did not run in the releases we are supposed to run :-) | 18:58 |
| elopio | alecu, ahhh, that's a different problem. | 18:58 |
| elopio | rmcbride, is working on updating those tests, and the idea is to have them being run by jenkins every time that the branch changes. | 18:58 |
| ralsina | alecu: but yes, that is a problem | 18:58 |
| alecu | elopio, cool. But we would anyway need the server running on every client version that we support | 18:59 |
| elopio | alecu, we should have that. That would be awesome. | 19:00 |
| alecu | elopio, surely would be awesome, but it looks a bit difficult to me. | 19:00 |
| ralsina | we could run the tests against the real server or a staging server or some server | 19:00 |
| ralsina | sure, the tests would take a day to run... | 19:00 |
| elopio | alecu, I think that's the next step. After having the integration and smoke tests automated, we need a QA lab with virtual machines for everything we support, and have jenkins shooting tests against those servers during the day. | 19:01 |
| alecu | elopio, because that means that we need to have the server tests running on Precise in the very near future. | 19:01 |
| dobey | alecu: having every version of the client we support, running against a test server, isn't that difficult now | 19:01 |
| dobey | alecu: just fire up X VMs, and have them all run the same test suite, with different versions of the client | 19:01 |
| elopio | if we don't have that, then we can't say we "support" those clients. | 19:02 |
| alecu | dobey, elopio, well, right now the client-server integration tests need to run in the same vm; facundobatista may correct me if I'm wrong. | 19:03 |
| ralsina | Once proxy is actually working, daily runs of the integration tests should be plenty to catch problems | 19:04 |
| dobey | alecu: the tests that are in ubuntuone-servers for that are special; if we had real CI with real clients talking to staging server all the time, we likely might not even need them | 19:04 |
| elopio | alecu, yes, I don't know about that. But wouldn't it require just to change the u1sdtool configuration as we do to test manually against staging? | 19:04 |
| elopio | and thinking it a second time, integration tests should be a little more independent from the server so we can test against a local server, against staging and against production. Then, they should be in a separate project. | 19:06 |
| elopio | rmcbride, what do you think about splitting u1-testing in u1-user-testing and u1-integration-testing? | 19:07 |
| rmcbride | elopio: I think that we should certainly have suites for both. I'm not entirely certain that we need to make them seperate projects, but we do need to cover both scopes, certainly | 19:07 |
| rmcbride | certainly | 19:08 |
| elopio | rmcbride, and what's the current status of the integration tests? I missed the weekly meeting, sorry. | 19:08 |
| dobey | elopio: i don't think "user-testing" is the right term for that | 19:09 |
| rmcbride | elopio: still in work. with holiday and such I still don't have an implementation. | 19:09 |
| rmcbride | acceptance testing is probably a better term (as it relies heavily on usability and user acceptability) | 19:09 |
| * rmcbride has never had his QA terminology completely correct though | 19:10 | |
| elopio | rmcbride, ok. If you need a hand let me know. | 19:10 |
| dobey | elopio: user testing, generally means putting a user in front of a computer, and having that user do things; at least, that's what it makes me think of when i see the term :) | 19:10 |
| dobey | user-tested; mom-approved | 19:10 |
| rmcbride | Or hooking a user up to wires and testing him | 19:10 |
| dobey | it's like kix | 19:10 |
| elopio | dobey, you name it, I'm also bad with names. It can be simulated-user-testing, ui-testing, acceptance-testing. | 19:10 |
| gatox | dobey, if we have a branch that don't show any lint issue in maverick, but it does in O... should we updated to use the new u1lint?? (for example this branch: https://code.launchpad.net/~alecu/ubuntuone-storage-protocol/timestamp-autofix-1-4/+merge/81758) | 19:10 |
| rmcbride | UNless we can actually start wiring some users up | 19:11 |
| dobey | gatox: not sure i understand? | 19:11 |
| elopio | we used to call them GUI-integration-testing. That's always a mess :) | 19:11 |
| dobey | gatox: stable branches can't depend on the new tools, no | 19:12 |
| gatox | dobey, ahhhhh ok..... so the answer is no | 19:12 |
| dobey | elopio: i am not sure we need to split that project up | 19:12 |
| dobey | gatox: also, you shouldn't vote on your own proposals :) | 19:12 |
| gatox | dobey, :S where did i do that? | 19:13 |
| rmcbride | elopio: I think there's room in the u1-testing project layout for testing in both contexts. | 19:13 |
| dobey | gatox: you voted "needs fixing" | 19:13 |
| dobey | oh, that's alecu's proposal | 19:13 |
| elopio | dobey, if we don't split them, then we could use the same bash scripts to configure the things that don't belong to our user interface. That might be good. | 19:13 |
| dobey | nevermind :) | 19:13 |
| gatox | dobey, yes | 19:13 |
| dobey | i thought that was your proposal | 19:13 |
| elopio | so we at least need a new level in the u1-testing hierarchy. | 19:14 |
| * rmcbride nods | 19:14 | |
| elopio | I'm going to eat now, in motorcycle \m/ Finally there's real sun in Costa Rica. | 19:16 |
| rmcbride | buen provecho | 19:17 |
| alecu | elopio, have a nice time! | 19:17 |
| alecu | elopio, rmcbride: thanks for your time! | 19:17 |
| ralsina | dobey: is the scope of the devtools package defined somewhere? | 19:18 |
| elopio | so, to wrap up my revised opinion: u1-client with unit tests that fake the proxies. move facundobatista's tests from u1-servers to u1-testing/integration, add scripts to set up proxies to u1-testing/scripts, have jenkins run simulated-users and integration tests with and without the proxies, for the 3000 platforms we support :) | 19:18 |
| ralsina | dobey: because, practically speaking, we need a place to put this :-) | 19:18 |
| alecu | gatox, to answer your question: u1-storage-protocol and u1-client: 1-4=M 1-6=N 2-0=O | 19:18 |
| elopio | rmcbride, alecu, thanks. bbl. | 19:18 |
| gatox | alecu, ok! thanks | 19:19 |
| dobey | ralsina: no, but i guess we should do that | 19:19 |
| ralsina | dobey: yes, we should | 19:19 |
| alecu | gatox, and sso-client: 1-0=M 1-2=N 1-4=O | 19:19 |
| dobey | ralsina: i know we do. but there's a difference between putting some new stuff in there, and tightly integrating it with it, as well | 19:19 |
| alecu | rye, duanedesign: I'm about to edit an article in the FAQ: https://one.ubuntu.com/help/faq/how-can-i-test-the-latest-ubuntu-one-builds/ | 19:22 |
| alecu | rye, duanedesign: what's the markup used? | 19:22 |
| duanedesign | alecu: rst | 19:24 |
| alecu | duanedesign, thanks! | 19:24 |
| duanedesign | alecu: http://rst.ninjs.org/ | 19:24 |
| duanedesign | that might be useful | 19:24 |
| alecu | duanedesign, I found that some pages (like the one above) have messed up the command line interactions. | 19:24 |
| gatox | duanedesign, cool | 19:25 |
| duanedesign | hymmmm | 19:25 |
| alecu | duanedesign, it looks like every line is pasted together. | 19:25 |
| alecu | duanedesign, I'm fixing that page, but just wanted to let you guys know about that. | 19:25 |
| duanedesign | thank you | 19:26 |
| alecu | duanedesign, I've fixed it like this: https://one.ubuntu.com/help/faq/how-can-i-test-the-latest-ubuntu-one-builds/ | 19:30 |
| alecu | duanedesign, it still looks a bit ugly; probably a better css style is needed for what the users should type on the command line | 19:31 |
| ralsina | EOD for me, since I seem to have missed have lunch. | 19:46 |
| gatox | ralsina, ack... | 19:48 |
| gatox | bye!! | 19:48 |
| gatox | ralsina, and have a great lunch :P | 19:48 |
| gatox | alecu, ping | 19:49 |
| gatox | alecu, this branch has conflicts: https://code.launchpad.net/~alecu/ubuntuone-storage-protocol/timestamp-autofix-2-0/+merge/81792 (really this time :P) | 19:50 |
| alecu | gatox, pong | 19:50 |
| alecu | gatox, it truly does, thanks! :-) | 19:50 |
| gatox | alecu, i'll install the vms tonight in order to test your branches | 19:54 |
| gatox | alecu, so i'll resume the reviews tomorrow | 19:54 |
| alecu | gatox, cool, thanks a bunch | 19:54 |
| gatox | alecu, no problem! | 19:54 |
| gatox | eod for me.... see you tomorrow people! | 20:07 |
| === gatox is now known as gatox_away | ||
| dobey | cheers all | 21:41 |
| === gatox_away is now known as gatox | ||
| alecu | bye all! | 22:12 |
| rmcbride | later all | 22:41 |
| === eu is now known as Guest29393 | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!