*** coolsvap|away is now known as coolsvap | 03:37 | |
*** mrmartin has joined #storyboard | 05:29 | |
*** mrmartin has quit IRC | 06:01 | |
*** mrmartin has joined #storyboard | 06:03 | |
*** mrmartin has quit IRC | 06:08 | |
*** jtomasek has joined #storyboard | 07:28 | |
*** coolsvap is now known as coolsvap|away | 09:56 | |
Zara | morning, storyboard! | 10:09 |
---|---|---|
SotK | morning! | 10:15 |
Zara | I think the characters issue strikes again: https://storyboard.openstack.org/#!/story/2000458 | 11:01 |
Zara | though I can't see what's strange about it | 11:02 |
Zara | seems like the same behaviour, though | 11:02 |
Zara | well, the list does eventually load in this case, but it takes a long time, and there's a prompt to debug/stop script on the description page | 11:04 |
SotK | weird, it seems ok (though slow) in chrome | 11:06 |
Zara | I get errors in the web console | 11:06 |
Zara | rror: [$sanitize:badparse] The sanitizer was unable to parse the following block of html: | 11:07 |
Zara | the description is different in the web console | 11:08 |
Zara | http://paste.openstack.org/show/483730/ | 11:09 |
Zara | so that doesn't match the description in the story | 11:10 |
SotK | does that happen if you refresh the story page? | 11:10 |
Zara | hm, seems not | 11:11 |
Zara | still getting the error | 11:11 |
Zara | sorry, an error | 11:11 |
Zara | but a different one | 11:11 |
Zara | Error: [$rootScope:inprog] $digest already in progress | 11:12 |
Zara | http://errors.angularjs.org/1.3.13/$rootScope/inprog?p0=%24digest | 11:12 |
Zara | I wonder what's going on, since that makes it sound like a more general bug, but it seems to only affects that story? | 11:18 |
Zara | *affect | 11:18 |
SotK | idk, pretty weird | 11:18 |
SotK | I don't know what would cause that | 11:18 |
Zara | still going with my 'sb doesn't like some syntax' theory but hard to narrow down. | 11:19 |
SotK | so the "db is really slow" issue is caused by the session hook | 11:26 |
Zara | yay, you found the cause? :D | 11:28 |
Zara | I wanted to look at email templates again today. currently sidetracked debugging thing above (since that will affect the jenkins job builder team) | 11:30 |
SotK | I think so, though I don't really know why it didn't take ages with query.count() separate from the query which retrieves the records | 11:32 |
SotK | I suspect there is some sqlalchemy magic at work that I don't understand | 11:32 |
Zara | replicated story bug on my own instance by copying over title and description | 11:33 |
Zara | okay the problem is somewhere in the block of text after the bit starting 'Moving the config check...' | 11:39 |
Zara | so that's: http://paste.openstack.org/show/483740/ | 11:45 |
Zara | I don't think it's an issue with a character now; I do think that some of the syntax is making storyboard interpret the subsequent code differently (maybe trying to run it? :/) | 11:50 |
Zara | copying in that whole block causes issue. copying in just first line seems to slow it down a little, but not lots (seems to get worse with each subsequent line). copying in all but first line has no effect. | 11:53 |
Zara | okay, so it dislikes allow_empty_variables = self.config \ | 11:56 |
Zara | but only when it's indented | 11:56 |
Zara | any idea what that might signify? | 11:57 |
SotK | what if it's indented by only four spaces? | 11:59 |
Zara | yeah, that's the indentation that seems to affect it | 12:00 |
SotK | I guess something about it causes the markdown parser to be slow | 12:00 |
Zara | and loop, it seems (in iceweasel anyway) I wondered but didn't know if the markdown parser worked when the save or preview button wasn't clicked | 12:01 |
Zara | my guess is that the markdown parser is reading it as an instruction (I'd think it was just confused but the other lines are similar syntactically and don't cause the same problem) | 12:03 |
SotK | what happens if you delete the \ from the end? | 12:04 |
Zara | it still dislikes it | 12:04 |
SotK | :S | 12:05 |
Zara | okay, swapping first line for just allow_empty_variables still causes issue | 12:09 |
Zara | swapping it for allow_fluffy_cats doesn't | 12:09 |
Zara | so that's the instruction that's making things go haywire | 12:09 |
Zara | ooh, but it doesn't like allow_fluffy_variables | 12:10 |
Zara | allow_variables doesn't trigger it, fluffy_variables doesn't trigger it, sleepy_fluffy_variables does. suggests it may be anything of form x_x_variables | 12:12 |
Zara | actually, also getting one without 'variables', so maybe it's to do with the length of the strings between underscores too | 12:15 |
Zara | or total length? | 12:15 |
paulsherwood | ah, ok | 12:17 |
paulsherwood | oops ww | 12:19 |
Zara | heh. :) | 12:19 |
Zara | as for this, I think I've got it! | 12:20 |
Zara | it's because the first string in the markdown description is too long | 12:20 |
Zara | where strings are separated by spaces | 12:20 |
Zara | if you indent four spaces and then start typing the alphabet, you can see the effect in realtime | 12:39 |
Zara | have commented on the affected story, anyway | 12:39 |
Zara | right, back to email templates! we're so close! I might have to ask for help getting them displaying urls; there was a mention of pecan and I'm still confused by it, though I'll look for a while before bothering anyone else. | 12:50 |
Zara | ack, I have 51 emails automatically sent from my test instance in the process of debugging that story. | 12:51 |
* pedroalvarez wonders if sending diffs for the changes in the description makes sense | 12:53 | |
pedroalvarez | @Line number for context | 12:55 |
pedroalvarez | Things that haven't changed | 12:55 |
pedroalvarez | - Thing that has been deleted | 12:55 |
pedroalvarez | + Think That has been added | 12:55 |
pedroalvarez | Another thing that hasn't been modified | 12:55 |
pedroalvarez | @ | 12:55 |
pedroalvarez | (just a thought) | 12:55 |
Zara | sounds interesting. I don't know that line number would help, since I don't think those are displayed on the story detail pages. the current template leaves the user to infer the difference between the old and new description, by posting 'from: {old description} to: {new description}' it'd be nice to have a future version of the template that narrowed it down, though. | 12:59 |
Zara | I'd like to improve that but I'm okay with the current implementation as the first step, provided it links to the story. (right now the user comment template and task deleted template are my priorities, because we have nothing there yet.) | 13:01 |
SotK | I've finally reached the bottom of the API slowness hole | 13:02 |
Zara | 'task deleted' seems a bit odd; I just get an error in the worker instead of a 'template not found' | 13:02 |
Zara | \o/ | 13:02 |
Zara | excellent | 13:02 |
Zara | wait, maybe I should hang on until you say what it is :P | 13:02 |
SotK | for some insane reason, it decides that all 682 (or however many stories you have) stories have been changed and need writing back to the database | 13:03 |
Zara | eep | 13:03 |
Zara | hm, okay, so when I comment on a story, I get an odd message from the worker | 13:05 |
Zara | http://paste.openstack.org/show/483753/ | 13:05 |
SotK | o.O | 13:06 |
Zara | and deleting a task gives me: http://paste.openstack.org/show/483756/ | 13:07 |
Zara | hm, actually, I also get the first one when I edit a description (though maybe not always) | 13:10 |
Zara | so maybe the problem here is more that commenting isn't having any effect on the worker | 13:10 |
Zara | still coming up in recent events, though | 13:10 |
SotK | aha, the insane reason turns out to be that summarize_task_statuses makes the objects "dirty" because it deletes some attributes and adds a new one | 13:32 |
SotK | argh | 13:32 |
Zara | so the lack of a TemplateNotFound for comments makes me think it's not trying to build the email | 13:49 |
Zara | if resource not in ['task', 'project_group', 'project', 'story', | 13:51 |
Zara | 'branch', 'milestone', 'tag']: | 13:51 |
Zara | return | 13:51 |
Zara | I wonder if that's why. | 13:51 |
Zara | hm, added 'comment', no dice | 13:57 |
SotK | oh, its the meeting in an hour! | 14:00 |
Zara | \o/ | 14:02 |
Zara | grr, having no luck with these mysterious user comments | 14:04 |
* SotK has no ideas about it | 14:04 | |
SotK | I'll look when I finish fighting stories | 14:04 |
Zara | heh, thanks | 14:06 |
Zara | hm, this also suggests they're treated differently: | 14:10 |
Zara | def handle_timeline_events(self, session, event, author, subscribers): | 14:10 |
Zara | for user_id in subscribers: | 14:10 |
Zara | if event.event_type == 'user_comment': | 14:10 |
Zara | event_info = json.dumps( | 14:10 |
Zara | self.resolve_comments(session=session, event=event) | 14:10 |
Zara | ) | 14:10 |
Zara | else: | 14:10 |
Zara | event_info = event.event_info | 14:10 |
SotK | heh, with a fix to stop stories becoming dirty, GETting 100 stories takes 364ms on storyboard.sotk.co.uk... on storyboard.o.o it takes 1.65s | 14:13 |
Zara | \o/ | 14:13 |
Zara | yes please! | 14:13 |
pedroalvarez | that is awesome | 14:15 |
* SotK pokes around to see if it can get any quicker | 14:16 | |
pedroalvarez | btw, you should compare the times in the same host, they might have different hardware | 14:16 |
Zara | tbf making it four times faster sounds pretty nice. iirc storyboard.sotk.co.uk has worse hardware? | 14:16 |
SotK | pedroalvarez: I know, I was just being lazy with that :) | 14:17 |
pedroalvarez | Zara: I would hope it has worse hardware :) | 14:17 |
Zara | so would I but tbf I don't know xD | 14:18 |
SotK | ok, I stopped being lazy with my speed comparison | 14:40 |
SotK | on storyboard.sotk.co.uk, after fix: 338ms | 14:40 |
SotK | on storyboard.sotk.co.uk, before fix: 5.93s | 14:40 |
Zara | ooooooooooo~ | 14:41 |
pedroalvarez | see, it's even better than we thought | 14:41 |
Zara | so... nearly 20 times faster? | 14:41 |
Zara | 17.5 according to a calculator | 14:43 |
betherly | woah SotK :D thats so awesome | 14:51 |
Zara | :D | 14:56 |
Zara | meeting in 4 minutes! so exciting! | 14:56 |
*** davidlenwell has quit IRC | 14:57 | |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard: Speed up story GET requests https://review.openstack.org/266961 | 14:58 |
Zara | oooooooo | 14:58 |
* SotK imagines he broke some tests | 14:59 | |
*** davidlenwell has joined #storyboard | 14:59 | |
Zara | I'll wait a bit before reviewing, then. still very excited! | 14:59 |
* SotK did | 15:18 | |
Zara | \o/ | 15:20 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard: Speed up story GET requests https://review.openstack.org/266961 | 15:22 |
SotK | Tests pass now \o/ | 15:57 |
Zara | ooooooooooo again | 15:57 |
Zara | will test now | 15:57 |
Zara | fuelled by doughnut | 15:57 |
* SotK sends https://review.openstack.org/#/c/267002/ | 16:00 | |
Zara | hm, in testing I noticed that there are no project groups listed in search (there are for dropdown for quicknav) | 16:04 |
Zara | that's not related to the patch, just noticed it now | 16:05 |
Zara | hah, but now it makes boards seem much slower by comparison xD | 16:06 |
SotK | which search has no project groups? | 16:11 |
SotK | oh, the main search page | 16:12 |
Zara | yeah | 16:14 |
Zara | (or the header search after pressing enter, which I think links to the same in practice) | 16:14 |
Zara | having tested the GETs patch it seems fine. :) the only thing I'm wondering about is whether we'll encounter hidden issues with this after prolonged use that were avoided by SQLAlchemy flushing every story | 16:21 |
SotK | it still flushes everything that has changed, just now nothing has changed for no real reason | 16:22 |
Zara | heh, cool :) I'm always wary with anything that has such a great impact on performance, but really hope it's all fine and we can merge it ASAP | 16:24 |
SotK | oh, I just realised there is already an established icon for tasks (see the search page) | 16:30 |
SotK | and all this time I've been using that weird one | 16:30 |
Zara | I dislike that one tbh | 16:31 |
Zara | I think it look smore like a 'tag' icon | 16:31 |
Zara | because it's literally a tag | 16:31 |
SotK | it suggests tag to me too | 16:32 |
Zara | I've even seen it and thought 'oh hey tags *are* already searchable, I was wrong', only to then squint and realise it says 'tasks' | 16:32 |
Zara | so I'd prefer that to change to being the tags icon. | 16:33 |
pedroalvarez | reviewed: https://review.openstack.org/#/c/266961/2 | 16:35 |
pedroalvarez | sorry I'm not reviewing a lot of things this year, but I couldn't not review this one | 16:36 |
SotK | no worries pedroalvarez, thanks for the review! | 16:36 |
Zara | \o/ yeah, any help is appreciated. and it's only the start of the year! :D | 16:37 |
Zara | hm, fa-pencil-square is still used on admin page, otherwise I'd say we should use that | 16:38 |
Zara | fa-pencil-square-o, even | 16:38 |
Zara | for a tasks icon, I mean | 16:38 |
SotK | the icon I've been using is pretty terrible, anything would be an improvement :) | 16:40 |
Zara | I'd suggest we use a normal pencil for the admin page edit button, since everywhere else uses that. then we free up the other icon for tasks | 16:41 |
Zara | ( http://fortawesome.github.io/Font-Awesome/icon/pencil-square-o/ ) | 16:41 |
pedroalvarez | that's the icon we are using for edits buttons | 16:42 |
pedroalvarez | right? | 16:42 |
Zara | only for one that I can find | 16:42 |
Zara | so that's why I was thinking of swapping that one for the usual pencil. but I might have missed some elsewhere | 16:43 |
SotK | its used for editing tasks in stories too | 16:43 |
Zara | oh, so it is, wah | 16:43 |
pedroalvarez | hm.. what about checkboxes? | 16:50 |
pedroalvarez | tasks are something that when you do, you tick a check box, right :) | 16:51 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard-webclient: Show Project Groups on search page https://review.openstack.org/267027 | 16:52 |
pedroalvarez | ( check-circle-o or check-square-o ) | 16:55 |
* SotK likes check-square-o | 16:57 | |
pedroalvarez | just an idea, I didn't find anything else interesting | 16:58 |
Zara | yeah, I think I liked the pencil-square because it was reminiscent of a checkbox without being one (concerned that using it as tasks icon might make users think they can check/uncheck it) | 17:00 |
Zara | anyone got a moment for looking at email comments weirdness | 17:05 |
SotK | oh yeah, I forgot I was going to peek at that :) | 17:05 |
Zara | :P well tbf you were busy making storyboard SEVENTEEN POINT FIVE TIMES FASTER | 17:06 |
Zara | I'm going to say that whenever I can | 17:07 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard-webclient: Improve "add card" UI https://review.openstack.org/267002 | 17:07 |
Zara | anyone opposed to me merging speedups? was going to hold off 'til the morning so we'd be around to fix anything that might go wrong, or in case anyone suddenly realised anything in the interim | 17:16 |
Zara | but maybe that's going too far. | 17:16 |
* pedroalvarez reviews a couple of things more that are ready to be merged | 17:17 | |
pedroalvarez | if you ask me my answer is " merge all the things" | 17:18 |
Zara | I like to stagger it a bit because we don't have as many eyes on the code (glad I waited with the unlimited pagination patch, hehe), but this one has been tested on production data tbf | 17:19 |
Zara | okay, people want it merged and we can revert if necessary | 17:19 |
Zara | here goes... | 17:19 |
pedroalvarez | yay! | 17:19 |
pedroalvarez | brave Zara is brave! | 17:20 |
SotK | \o/ | 17:20 |
Zara | I'm less worried with this because we've all tested it in different ways with different amounts of data | 17:21 |
SotK | thanks Zara :) | 17:21 |
Zara | but stilllll... speedups always scare me xD | 17:21 |
SotK | I wonder why openstackgerrit doesn't notify us of things sometimes | 17:27 |
openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard-webclient: Show Project Groups on search page https://review.openstack.org/267027 | 17:35 |
Zara | night-night, storyboard! :) | 17:59 |
openstackgerrit | Merged openstack-infra/storyboard-webclient: Improve "add card" UI https://review.openstack.org/267002 | 18:03 |
*** jjardon has quit IRC | 18:08 | |
*** betherly has quit IRC | 18:08 | |
*** jjardon has joined #storyboard | 18:23 | |
*** betherly has joined #storyboard | 18:23 | |
*** davidlenwell has quit IRC | 19:57 | |
*** davidlenwell has joined #storyboard | 19:58 | |
*** jtomasek has quit IRC | 21:32 | |
* SotK realises too late he meant to reduce the live search results count from 500 to 50 | 23:29 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!