| *** openstackgerrit_ has joined #storyboard | 06:51 | |
| *** jtomasek has quit IRC | 07:10 | |
| *** ttx has joined #storyboard | 08:09 | |
| *** jtomasek has joined #storyboard | 09:37 | |
| SotK | good morning! | 10:02 |
|---|---|---|
| pedroalvarez | morning! | 10:12 |
| Zara | morning! | 10:16 |
| *** openstackgerrit_ has quit IRC | 10:23 | |
| * pedroalvarez has noticed this weekend that the 401 never happens from his phone | 10:30 | |
| * Zara tweaks 'about' page so it reads: "Storyboard: For Phones Only" | 10:31 | |
| Zara | problem solved =D | 10:31 |
| pedroalvarez | hehe | 10:31 |
| pedroalvarez | what I wanted to point out, is that it works for some environments.. | 10:47 |
| Zara | yeah, it's interesting. I wonder whether that means it's more or less broken on phones | 10:48 |
| Zara | ie: whether the code that says 'log out this user' isn't working, or whether the code which says 'log this user back in' is working | 10:49 |
| pedroalvarez | well, for an user it looks like it's working properly, which is good | 10:52 |
| pedroalvarez | I noted also that when you save the preferences, the green "pop up" doesn't appear until you go to another page.. | 10:52 |
| pedroalvarez | this might be related | 10:53 |
| pedroalvarez | who knows! but one day we will know for sure :) | 10:53 |
| SotK | hmm, I wonder if the mobile browser is handling the bit that refreshes the token differently for some reason | 10:54 |
| SotK | (ie. actually doing it before giving up on the request) | 10:54 |
| Zara | is there a way of seeing the relevant code for the mobile browser? | 10:58 |
| Zara | (idk what browser it is) | 10:58 |
| pedroalvarez | this was on latest chrome for android | 10:59 |
| pedroalvarez | oh god, I just tried again and 401s happened.. | 11:00 |
| pedroalvarez | I promise you that I wasn't lying .. :( | 11:01 |
| Zara | :0 | 11:12 |
| openstackgerrit | Merged openstack-infra/storyboard-webclient: Add a preview button for story descriptions https://review.openstack.org/252055 | 11:34 |
| *** jtomasek has quit IRC | 11:42 | |
| *** jtomasek has joined #storyboard | 11:45 | |
| *** jtomasek has quit IRC | 11:46 | |
| *** jtomasek has joined #storyboard | 11:48 | |
| openstackgerrit | Pedro Alvarez Piedehierro proposed openstack-infra/storyboard-webclient: Add templates for tags added/deleted subscription events https://review.openstack.org/251605 | 12:35 |
| openstackgerrit | Pedro Alvarez Piedehierro proposed openstack-infra/storyboard-webclient: Add checkboxes for tag events on preferences page https://review.openstack.org/254162 | 12:35 |
| pedroalvarez | adding the checkboxes was really easy to be honest | 12:39 |
| *** tlbr has quit IRC | 13:09 | |
| *** tlbr has joined #storyboard | 13:50 | |
| *** ttx has quit IRC | 14:23 | |
| *** ttx has joined #storyboard | 14:24 | |
| Zara | ack, just noticed that the worklists/boards page currently does one GET per worklist, one GET per board | 14:26 |
| Zara | I'm guessing no one user is going to have a million billion worklists/boards, but it's still an issue. | 14:27 |
| pedroalvarez | is this because it would be more efficient? | 14:37 |
| Zara | my guess is it was the easiest implementation, but it'll be slow/won't scale well. I'd've thought it wouldn't need to scale loads but I might not know about a use-case. | 14:38 |
| SotK | yeah, its very much not optimised | 14:39 |
| SotK | its worse than one GET per worklist too | 14:40 |
| Zara | ooh? =D | 14:41 |
| SotK | for each board it does something like `GET /v1/boards/:id`, `GET /v1/worklists?board_id=:id`, then for each worklist that is returned `GET /v1/worklists/:id/items` | 14:42 |
| Zara | ah | 14:43 |
| SotK | so 2 GETs per worklist (worklist and the items), and "2 + number of lanes" GETs per board | 14:43 |
| * SotK has plans to make it less awful | 14:43 | |
| Zara | lol, for a second I read that as 'SotK has no plans to make it less awful' | 14:43 |
| Zara | I was like, '... good to know...' | 14:44 |
| Zara | but it turns out it *is* good to know, so thanks :) | 14:44 |
| SotK | :D | 14:44 |
| Zara | permissions patch is merging... NOW! | 14:55 |
| Zara | well, I've +1d it, anyway | 14:56 |
| Zara | so hopefully pedro's patches won't be blocked for much longer | 14:56 |
| pedroalvarez | yay!!!!!!!!1 | 15:00 |
| Zara | :P I'm pretty sleepy today, but I'll try to test the docs for setting up notifications tomorrow. | 15:00 |
| pedroalvarez | I'll change that 49 to a 50 and that 50 to a 51 :) | 15:00 |
| SotK | Zara: thanks :) | 15:01 |
| openstackgerrit | Merged openstack-infra/storyboard: Implement complex permissions for worklists and boards https://review.openstack.org/235472 | 15:04 |
| SotK | \o/ | 15:04 |
| Zara | yeah, we need the resource caching :/ a story detail page on my instance is doing about 25 more GETs than it was before. | 15:28 |
| openstackgerrit | Pedro Alvarez Piedehierro proposed openstack-infra/storyboard: Add story_title to tag events. https://review.openstack.org/251566 | 15:33 |
| *** jtomasek has quit IRC | 16:46 | |
| Zara | hmm, ugly bug in storyboard | 16:52 |
| Zara | https://storyboard.openstack.org/#!/story/2000440 | 16:53 |
| Zara | have replicated it. | 16:56 |
| Zara | my guess would be that the title of a story after #25 is getting misinterpreted and we need to sanitise better, but that's just a guess | 16:58 |
| pedroalvarez | wow, scary bug | 17:01 |
| SotK | oh dear | 17:02 |
| SotK | witness the horror: https://storyboard.openstack.org/api/v1/stories?limit=10&offset=30&project_id=679&status=active | 17:07 |
| Zara | tried putting all the titles that looked like they had interesting characters into a project in my instance, didn't cause the problem | 17:07 |
| Zara | I daresay the person with the actual database data can do a better job debugging xD | 17:09 |
| * SotK is trying to figure it out | 17:11 | |
| Zara | so story 31 or 31 or something? | 17:11 |
| Zara | I went with limit 1 and then increased the offset... | 17:12 |
| Zara | jobs should not be restarted indefinitely | 17:15 |
| Zara | ^I think it's that one | 17:15 |
| Zara | it's freaking out about the description | 17:15 |
| Zara | https://storyboard.openstack.org/#!/story/1260910 <- stop script, then click edit | 17:16 |
| Zara | and ye shall see | 17:16 |
| persia | As a data point, that last URL loads for me without stopping any scripts | 17:16 |
| SotK | does the content display properly too? | 17:18 |
| persia | yes | 17:18 |
| persia | Same behaviour logged-in or logged-out, but I thought I'd note it since it seems different from other's experience. | 17:19 |
| Zara | I'll try it in chromium | 17:19 |
| Zara | iceweasel doesn't like it | 17:19 |
| SotK | chromium doesn't either | 17:19 |
| Zara | yup | 17:20 |
| pedroalvarez | not valid xml | 17:21 |
| Zara | clicking edit allows us to see the original description, unrecognised characters and all | 17:21 |
| Zara | we could edit it, or change the code (or copy it, edit it, and change the code later-- though we need it to be an accurate copy) | 17:22 |
| Zara | well, I'm assuming it'll let us edit it. | 17:22 |
| * SotK notices its working in iceweasel on jessie for him | 17:23 | |
| Zara | are you also getting the page break in the zuul stories list | 17:23 |
| Zara | ? | 17:23 |
| SotK | not in iceweasel on jessie | 17:25 |
| SotK | seems like its coping with the error caused | 17:25 |
| SotK | which seems to be angular's html sanitiser tripping up on the story description | 17:26 |
| persia | SotK: Your experience with iceweasel/jessie matches mine, so I suspect I can't help more. | 17:27 |
| persia | Maybe bits of angular don't run on some browsers? | 17:27 |
| Zara | yeah, or the browser itself is compensating on jessie? | 17:28 |
| SotK | actually, having said that I don't see anything in the console when loading the story detail page | 17:28 |
| SotK | so I guess the sanitisation troubles are coming from elsewhere | 17:29 |
| * pedroalvarez finds unicode on the description | 17:29 | |
| Zara | yeah, it's funky. | 17:29 |
| SotK | I guess we need to handle the unicode on the API side, since it isn't valid xml | 17:30 |
| Zara | anyone want to/agree to try editing it for now? copy the unicode characters somewhere so we can put them back in later, and remove them from the description in the meantime? assuming that getting the API to handle unicode is more than a two-minute job, anyway | 17:31 |
| pedroalvarez | they are escape unicode characters I think | 17:32 |
| pedroalvarez | so, nothing to show | 17:32 |
| pedroalvarez | this would be the description unicode free: http://paste.openstack.org/show/481058/ | 17:34 |
| persia | Um. Except for a few codepoints listed at http://www.w3.org/TR/unicode-xml/#Suitable , the rest of Unicode *is* valid XML. | 17:34 |
| Zara | okay, I can't edit it because attempting to freezes my browser. xD persia or SotK, it's all yours. | 17:35 |
| * persia would be annoyed not to be able to copy "12月 22" from a terminal window when showing a log file | 17:35 | |
| persia | Zara: Use a browser in an older VM (e.g. jessie). I'm using rekonq on trusty, for example. | 17:36 |
| * SotK wonders if persia can load https://storyboard.openstack.org/api/v1/stories?limit=1&offset=32&project_id=679&status=active successfully | 17:36 | |
| SotK | (in browser) | 17:36 |
| persia | SotK: That reports "This page contains the following errors:" | 17:37 |
| persia | Using an inspector, the entire error seems to be valid XML. | 17:37 |
| persia | But it *is* an error report. | 17:37 |
| * SotK gets "XML Parsing Error: not well-formed" | 17:38 | |
| persia | SotK: In jessie/iceweasel? | 17:39 |
| SotK | yep | 17:39 |
| persia | http://paste.openstack.org/show/481059/ | 17:40 |
| pedroalvarez | well, different browsers, different errors, but the error is clear "error on line 4 at column 66: PCDATA invalid Char value 27" | 17:41 |
| persia | pedroalvarez: Two issues: #1: the loose escape character. #2: that the presentation of this breaks some clients. | 17:42 |
| Zara | I could make a brand new vm with a different browser, or someone for whom the page works could just edit the story for now (copying the characters to use them for testing purposes) | 17:42 |
| persia | It should work for folk not me, or even I can't use it to collaborate :) | 17:43 |
| persia | Zara: As everything works for me, I have no idea what to change. | 17:43 |
| SotK | http://paste.openstack.org/show/481061/ is the offending part it seems | 17:45 |
| Zara | there's a bit further on, hang on | 17:46 |
| Zara | http://paste.openstack.org/show/481062/ | 17:46 |
| Zara | :) | 17:46 |
| persia | Aha! | 17:46 |
| persia | So, the issue is that the ANSI control codes (or whatever folk are using these days) are transliterated in such a way that they appear to be XML literals. | 17:47 |
| persia | It's mostly about '[', which probably ought be encoded by the content sanitiser before being stored in the DB. | 17:47 |
| persia | And the DB should be cleaned. | 17:48 |
| Zara | anywhere there's a square bracket in those pastes, there's *secretly* a unicode character | 17:48 |
| persia | It isn't secret, especially. | 17:48 |
| persia | And, at least from the error above, it isn't even espeically unicode. | 17:48 |
| Zara | :P well, the pastebins don't display them. I like to pretend it's top-secret. | 17:48 |
| persia | It's valid ASCII. | 17:48 |
| persia | Just terminal escape codes (which are called "escape codes" for a reason) | 17:49 |
| persia | And if these cannot be safely stored in the DB and displayed, then nobody can copy&paste from a terminal. | 17:49 |
| persia | Which would be inconvenient. | 17:49 |
| persia | So, anyway, *this* story can be sanitised, but it would be good to check for other stories to see where else people copied text from their terminal. | 17:52 |
| persia | Because there's probably a fair number of stories in need of cleaning. | 17:52 |
| Zara | yeah, agreed; I think this one was only discovered because people use the project a lot | 17:54 |
| pedroalvarez | wow, I copied and pasted the description sanitized on my storyboard, and unicode happens again | 17:54 |
| Zara | longer term, storyboard needs to handle things correctly. | 17:54 |
| persia | pedroalvarez: When you say "unicode", what do you mean, precisely? Multibyte glyphs? | 17:54 |
| persia | wide characters? | 17:55 |
| pedroalvarez | there is a 0x1b char jjust before every '[' | 17:55 |
| pedroalvarez | on the returnted xml from the api | 17:55 |
| persia | Right. That's the "ESC" character from ASCII, used in terminal escape codes. | 17:55 |
| persia | https://en.wikipedia.org/wiki/ANSI_escape_code | 17:56 |
| Zara | okay, attempting to save sanitised version | 17:57 |
| pedroalvarez | what I mean is that if you create a story, without these "ESC" chars, they will appear again, so my guess is that the description doesn't have any unicode char, but the api is creating them | 17:57 |
| persia | Unlikely, in that specific structure, with that specific content. | 17:58 |
| Zara | I think it might have not been santized fully. I had weird issues editing it until I deleted the square brackets, too. | 17:58 |
| persia | Possible, but unlikely. | 17:58 |
| Zara | I'm not sure why but it seems to work, now, anyway. | 17:58 |
| Zara | maybe a caching thing, idk | 17:58 |
| Zara | I have the old unicode stuff saved if we need another copy | 17:59 |
| *** b3rnard0 is now known as b3rnard0_away | 18:00 | |
| pedroalvarez | this is totally reproducible | 18:00 |
| persia | Please don't call this "unicode": it is not a unicode-relarted problem. | 18:01 |
| persia | "ANSI" is probably the most equivalent term, if you want a one-word entry to specify the standard under which these codes exist. | 18:01 |
| Zara | okay. the zuul team should be able to see their projects, now, anyway. | 18:04 |
| persia | Cool! | 18:06 |
| Zara | aw, I'm so happy, we got to the bottom of a bug! :) in record time! | 18:06 |
| pedroalvarez | well.. this can happen again :) | 18:07 |
| pedroalvarez | in the meantime I've put 月 on a story, and that works good | 18:07 |
| Zara | yeah, but at least we know why ~o~ | 18:07 |
| Zara | ~you can't take my celebration away from me~ | 18:07 |
| pedroalvarez | haha | 18:07 |
| Zara | 'night, all! :) | 18:08 |
| persia | 月should be fine: it's nice printable unicode. ANSI codes are explcitly non-printable, which causes the issue. | 18:08 |
| SotK | Zara: night :) | 18:08 |
| SotK | persia: how should the ANSI codes be stored in the database? | 18:09 |
| persia | SotK: I'd recommend dropping them entirely. | 18:09 |
| persia | They usually indicate things like "move the cursor back over what was just written to overwrite it" or "change the color", neither of which is often interesting in log files. | 18:10 |
| persia | (as the log should be colorised by the log renderer, not in the log, and the overstrike is only interesting in realtime) | 18:10 |
| SotK | aha, that explains why I couldn't see any issue when looking at the database itself :) | 18:19 |
| *** b3rnard0_away is now known as b3rnard0 | 18:35 | |
| *** jtomasek has joined #storyboard | 19:26 | |
| *** jtomasek has quit IRC | 19:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!