07:34:43 #startmeeting 07:34:43 Meeting started Wed Aug 3 07:34:43 2011 UTC. The chair is dan_p. Information about MeetBot at http://wiki.debian.org/MeetBot. 07:34:43 Useful Commands: #action #agreed #help #info #idea #link #topic. 07:34:49 #topic Meeting attendees 07:34:57 Please put #info in from of your name 07:35:02 #info dan_p is Dan Poltawski - LUNS Ltd, UK 07:35:11 #info dobedobedoh is Andrew Nicols - LUNS Ltd, UK 07:35:17 #info hughdavenport is Hugh Davenport, Catalyst IT 07:35:18 #info anitsirk: Kristina Hoeppner - Catalyst IT, Wellington, NZ 07:35:21 #info richardm is Richard Mansfield - Catalyst IT Wgtn 07:36:25 Some housekeeping whilst waiting for any other latecomers.. 07:36:41 please say ".." on a line by itself if you are typing a long multi-line piece to indicate you've finished 07:36:47 .. 07:36:51 #info iarenaza is Iñaki Arenaza - Mondragon Unibertsitatea 07:37:48 Ok, I think we are in for a short and sweet meeting today, so i'll move on ;) 07:37:58 #topic Items from previous meeting 07:38:03 #info fmarier_ to update the bug status page with new "milestone" usage guideline 07:38:12 In the absenece of fmarier I checked - and don't believe this has been completed https://wiki.mahara.org/index.php/Developer_Area/Bug_Status 07:38:31 So I propose to carry this action forward to the next meeting 07:38:54 #agree 07:38:58 #agree 07:39:01 #agree 07:39:04 #agree 07:39:08 #action fmarier_ to update the bug status page with new "milestone" usage guidelines 07:39:24 #info thomaswbell88 to come up with ideas for mahara.org badges and post on the wiki under Specifications in Development 07:39:32 In the absence of thomaswbell88 I checked and found the wiki page WIP: 07:39:36 https://wiki.mahara.org/index.php/Developer_Area/Specifications_in_Development/Forum_Badges 07:40:29 Any comments on this without thomaswbell88 present? 07:40:46 (Otherwise I propose to carry it forward to the next meeting..) 07:41:03 * iarenaza having a look at it right now 07:41:26 sounds good, probably needs a bit more technical detail before we go and implement it 07:41:49 yeh page looks good, but will need filling out a bit 07:42:00 and as richardm says a bit of tech detail 07:42:36 #info Like the work in progress in wiki, but needs more technical detail to implement 07:42:44 Yup 07:43:34 Ok, well i'll carry that action forward 07:43:40 #action thomaswbell88 to come up with ideas for mahara.org badges and post on the wiki under Specifications in Development 07:43:52 #topic Unit testing (dobedobedoh) 07:44:27 As the agenda page says, I've updated Mjollnir`'s phpunit branch from way back 07:44:48 I've put it up git.luns.net.uk (https://git.luns.net.uk/?p=mahara.git;a=shortlog;h=refs/heads/phpunit) 07:44:52 if anyone wants to see 07:45:02 I'd like to propose that we try and get it in to Mahara 1.5 07:45:20 and I was hoping for some guidance on the most appropriate way of merging it 07:45:38 There are about 8 commits off master which are a mixture of Mjollnir`'s and mine 07:45:49 and several are missing Signed-off-by etc 07:45:50 #info dobedobedoh has updated Mjollnir`'s phpunit branch with the intention of merging into 1.5 07:46:49 So I also wanted to raise the question as to the best way of merging works which don't meet our current guidelines 07:47:08 dobedobedoh: i think it's not a problem to retain authorship, i pushed a couple of patches that weren't mine to gerrit recently 07:47:14 And to ask whether we should squash all of the commits down into one etc 07:47:52 depends on whether it is one big change or not 07:47:57 dobedobedoh: whatever you think will make them easiest to review i suppose 07:48:01 Okay dokay 07:48:02 it is easier to review if it is smaller chunks 07:48:10 but not too small :) 07:48:24 you could just push what you have and we can work from there 07:48:39 Okay. I'll tidy it up a bit first as there are a few I'd like to squash 07:48:48 dobedobedoh: Can we ask Mjollnir` to add the signed-off lines? (if that's a real problem) 07:49:14 gerrit will complain if you push stuff that has signedoffby someone else 07:49:22 you can change the author on the commit 07:49:39 the signedoffby is basically saying the person that pushed it is alright with it 07:49:48 hughdavenport: and if it has multiple signed-off lines, does it also complain? 07:49:49 but they may not be the original author 07:49:54 dobedobedoh can push them with his own signed-off, and Mjollnir` as author 07:50:16 iarenaza: not sure, but iirc it did it for me 07:50:26 iarenaza: it's happy with multiple signed-off lines i think 07:50:40 yeh could be that 07:50:40 That was all really -- I just wanted to raise the profile of the unit testing, and ask you to review the patches when they hit gerrit ;) 07:50:43 .. 07:50:44 dobedobedoh: are you proposing to acompany the code with some developer documentation? ;-) 07:50:46 just had to have you in it somewhere 07:50:48 It is happy with multiple signed off lines 07:51:11 dan_p: There's already a WIP at https://wiki.mahara.org/index.php/Developer_Area/Unit_Testing 07:51:31 But it's very much a WIP ;) 07:52:30 That's all I have 07:52:48 Please consider what tests you may want to write... 07:52:49 .. 07:53:24 #info dobedobedoh to update the code into reviewable chunks 07:53:26 #info Ruslan Kabalin - LUNS ltd., Lancaster, UK 07:54:31 dobedobedoh: I don't think there are any actions out of that, agree? 07:54:38 agree 07:54:45 except on me to push 07:55:01 ok will move on.. 07:55:02 #topic Next meeting 07:55:10 I think hugh had something 07:55:17 yep, i added it last minute 07:55:19 in the agenda 07:55:24 well last day/minute :P 07:55:27 doh 07:55:42 won't take long :) 07:55:43 #undo 07:55:43 Removing item from minutes: 07:56:03 #info Selenium/Unit tests with commits 07:56:24 right with this, I was thinking that it could be an idea for when we push a commit with a new feature 07:56:33 #topic Selenium/Unit tests with commits 07:56:45 that we also push along side it a unit test or a selenium test 07:56:57 that way we can test when something breaks later on 07:57:03 and we keep the tests up to date 07:57:21 and another thing is that we could possibly integrate one or both into jenkins 07:57:25 thoughts? 07:57:26 .. 07:57:29 #agree 07:57:31 #agree 07:57:47 #info Push a unit/selenium test with a new feature 07:57:52 I've tried this in the past... getting the selenium tests up to date 07:57:55 it will take a bit to get into the habit of doing, but I think it will be good in the long run 07:58:21 The difficulty I found was in getting them detailed enough to test the functionality I'd written 07:58:33 BUt I think that combined with phpunit tests, it shouldn't be a problem 07:58:33 .. 07:58:41 yeh that is the problem with any test suite really 07:58:45 combining does help 07:58:56 #idea hughdavenport proposes that developers push a selenium/unit test with a new feature 07:59:11 Selenium is a particular pain as it doesn't support state between tests or dependencies 07:59:21 yeh... 07:59:27 I have got this impression of Selenium being particularly 'fragile' 07:59:31 but with your new phpunit that will be good 07:59:54 dan_p: well it's basically a simulation of users clicking around, easily broken :) 07:59:57 There's also that... some things you end up testing with xpath, and then something changes in teh UI and it's a right pain to calculate the new xpath 08:00:02 basically any change could break it 08:00:28 they also take forever to run 08:00:40 And automating them with jenkins is potentially tricky 08:00:47 though not impossible 08:00:49 they do, which may mean that jenkins may not be the best 08:01:03 is it possible to have two jenkins runs, a quick and a slow? 08:01:06 I definitely agree they should be tested with jenkins though 08:01:21 hughdavenport: yes I think so 08:01:27 so we can get feedback on when it fails straightaway, and then later whether the tests pass 08:01:32 Some things can only be tested with UI tests such as selenium so we definately need them 08:01:58 hughdavenport: though, does that not sort of break the workflow of trying not to 'break the build', if you submit and carry on? 08:02:27 phpunit every commit, & selenium once or twice a day? 08:02:36 We already have a requirement to split out minaccept and jenkinsaccept in the Makefile 08:02:47 Ah I see 08:03:14 dobedobedoh: Are selenium tests less maintaince than detailed human instructons/testers.. 08:04:05 Debateable... they're more repeatable, but because of things like sometimes having to locate elements by xpath 08:04:15 they can take a lot of setting up 08:04:17 all this said, the tests shouldn't be taken as gospel, they are just useful for finding stuff that may break old stuff accidentaly 08:04:21 and then be broken very quickly 08:04:22 .. 08:04:41 I think they're worthwhile, but only when combined with phpunit 08:04:54 i would agree with that 08:04:59 and I think it's probably worth looking at some of the user contributed comonents 08:05:06 i like richardm's comment saying maybe phpunit every commit 08:05:12 e.g. I came across one which allowed you to store state between tests 08:05:15 then selenium less regularily 08:05:39 I think this is proably worth revisiing after phpunit is integrated? 08:05:51 yep, that sounds good 08:05:54 Or do we need to look at selenium integration now? 08:05:55 Yeah 08:06:02 Selenium is already integrated 08:06:09 or you mean with jenkins? 08:06:10 integration to jenkins 08:06:23 i think integration with jenkins can come later 08:06:43 we can integrate phpunit first as it is more useful for jenkins on a per commit as its faster 08:06:56 Are we bothered about things taking a while to test at present? There doesn't seem to be a /massive/ queue 08:07:07 but the whole point of my topic was just to get the mindset of thinking about writing tests as we go 08:07:24 hughdavenport: I like it :) 08:07:33 dobedobedoh: when we submit a large series of patches it would have to test them all 08:07:37 true 08:07:39 that woudl suck 08:07:40 when it would be better to test at the end 08:07:54 i think we do need integration with jenkins, but it should be pulled straight from the main repo, not gerrit 08:08:07 (for selenium i mean) 08:08:10 especially if there are multiple patchsets 08:08:17 ah, agreed 08:08:33 like richardm said, it would be better to test after merge, we can always revert (although ugly) 08:08:37 richardm: would you see that then generating a bug or alerting some other way? 08:08:39 but it's only a whatif case 08:09:05 dan_p: spamming the committer i guess 08:09:18 It'd ahve to track down /which/ commit broke things 08:09:18 #info richardm suggests intergating selenium with jenkins, but seperate from jenkins 08:09:36 separate from gerrit i meant 08:09:39 doh 08:09:42 hehe, integrate with jenkins but not with jenkins :P 08:09:46 #unfo 08:09:48 #undo 08:09:48 Removing item from minutes: 08:09:51 hehe 08:09:55 #info richardm suggests intergating selenium with jenkins, but seperate from gerrit 08:10:00 more coffee required 08:10:01 early morning dan_p? 08:10:06 We could have just two projects on jenkins 08:10:26 one for gerrit another for selenioum from master 08:10:33 * iarenaza is sorry, but has to leave (will read the chat logs later) 08:11:17 #action LUNS to investigate adding a new mahara integration project to run selenium tests 08:11:41 :) 08:12:47 Ok, shall we leave it at that? 08:12:56 sounds good 08:13:06 #topic Next meeting 08:13:23 I proposed on the wiki 01/09/11 - 17:30 UTC 08:13:36 Thursday evening UK time, Friday morning NZ 08:14:19 sounds good 08:14:21 I agree 08:14:22 i can make it earliest on sept. 5 08:14:50 but if there is only tech stuff as today, i can read the log ;-) 08:15:12 anitsirk: would be good to have you there 08:15:20 I can't make 5th, so new proposal 08:15:31 06/09/11 - 19:30 UTC 08:15:43 (sorry that previous time was supposed to be 19:30!) 08:16:02 fine by me 08:16:03 the 7th should work for me 08:16:03 I was gonna say! 08:16:06 lol, just saw that 08:16:08 WFM 08:16:19 :) 08:16:29 6th sounds fine 08:16:38 #info Next meeting agreed for 06/09/11 - 19:30 UTC 08:16:53 #idea dan_p to drink more coffee before future meetings 08:16:59 hehe 08:17:07 #undo 08:17:07 Removing item from minutes: 08:17:09 #topic Any other business 08:17:53 Going once.. 08:17:59 nothing from me 08:18:17 twice.. 08:18:20 nope. all good 08:18:47 Ok, lets wrap it up 08:18:58 thanks everyone 08:18:59 #endmeeting