16:01:49 <david-lyle> #startmeeting Horizon 16:01:50 <openstack> Meeting started Tue Apr 15 16:01:49 2014 UTC and is due to finish in 60 minutes. The chair is david-lyle. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:51 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:54 <openstack> The meeting name has been set to 'horizon' 16:01:56 <lsmola> hello 16:01:58 <david-lyle> Hello everyone 16:02:03 <jpich> Hello 16:02:04 <jcoufal> o/ hello everyone 16:02:06 <jrist> o/ 16:02:11 <tmazur> hello o/ 16:02:15 <santibaldassin> hi everyone 16:02:23 <akrivoka> hey 16:02:58 <david-lyle> #link https://wiki.openstack.org/wiki/Meetings/Horizon 16:03:10 <david-lyle> #topic: Release Status 16:03:54 <david-lyle> RC2 appears to be holding up with no blocking issues having been raised, this looks like our final RC 16:04:04 <david-lyle> \o/ 16:04:24 <akrivoka> awesome 16:04:50 <david-lyle> The last thing I need to do is post the release notes 16:05:18 <david-lyle> when I do that, hopefully today, I'll ask everyone to take a look and correct my mistakes 16:05:23 <jpich> Cool, looking forward to the patch! 16:05:58 <david-lyle> Aside, from that, I think we're focused on Juno 16:06:36 <jrist> well done, everyone 16:07:15 <david-lyle> The deadline for session proposals is five days away, so if there's something you feel needs to be discussed at the summit, get your proposal in 16:07:59 <david-lyle> #topic DocImpact tag (jpich) 16:08:05 * jcoufal will add one more session proposal focused on tuskar-ui 16:08:08 <jpich> Hey! 16:08:18 <david-lyle> jcoufal +1 16:08:21 <jcoufal> (sorry for jumping into your topic, jpich) :) 16:08:32 <jpich> I was chatting with someone on the docs team last week and realised we haven't been very good at using the DocImpact tag in Horizon (or I missed it!). It would be helpful to the docs team if we could include the tag at least for major changes and "procedure" changes - I'm guessing things around new configuration / upgrade / installation information. 16:08:37 <jpich> jcoufal: We're flexible here ;) 16:08:52 <jpich> I'm still a bit fuzzy on when it is appropriate to use it for us (so many changes impact the user experience in some way) but we can refine our approach to using the tag as we understand it better :) 16:09:09 <jpich> Somme additional reading: https://wiki.openstack.org/wiki/Documentation/DocImpact http://justwriteclick.com/2013/09/17/openstack-docimpact-flag-walk-through/ and I'm sure the folks at #openstack-doc are happy to answer any question, too 16:10:02 <tzumainn> jpich, one question 16:10:23 <tzumainn> looking through the baremetal example in the second link, it strikes me that a lot of those questions are exactly the same ones a reviewer would have 16:10:40 <david-lyle> other than internal bug fixes, don't all changes in the UI potential doc impact 16:10:52 <tzumainn> are reviewers expected to just ask those questions through the gerrit comment system? 16:10:58 <david-lyle> have potential that is 16:11:18 <tzumainn> or is there an expectation that the commit message have a certain level of self-documentation? 16:11:44 <jpich> tzumainn: I'm guessing in an ideal world these should be answered in the commit message and blueprint details, yes 16:12:15 <tzumainn> jpich, is that a standard - i.e., should reviewers -1 and ask for details on how to test a change? 16:12:16 <jpich> david-lyle: These are my thoughts as well, which is when it was clarified to me to care more about "procedure" changes rather than changes in the look 16:12:54 <jpich> david-lyle: I think we should start using it for major changes, the kind we self-document at the moment. Some of this should probably also live in the generic openstack guides? 16:13:10 <jpich> like settings, or how to use policies 16:13:29 <jpich> how to set up the dashboard to work with keystone v3 - probably these would have been great commits for DocImpact 16:13:44 <david-lyle> jpich, I see 16:14:21 <jpich> tzumainn: I think it'd be great to have that information but it hasn't been customary to block on it being missing. I'd lean toward strongly encouraging people to think about it and add that sort of info to the bug reports or blueprints, but not block on it... That'd be my thoughts on that 16:14:35 <tzumainn> jpich, okay, that sounds sensible enough - thanks! 16:14:58 <tzumainn> er, sorry, I now realize that I was re-directing away from your intended topic :P 16:15:04 <jpich> david-lyle: I'm still figuring this stuff out too :) 16:15:47 <david-lyle> jpich, thanks for raising, I think this a great thing to keep in mind. Even if it just means more complete commit messages for Horizon use. I think we've been flagging a bit on those lately 16:15:50 <jpich> tzumainn: It's documentish! I was bringing this up as a general "FYI, more things we should probably try to pay attention to" 16:16:25 * lsmola has to disappear soon today 16:16:35 <amotoki_> hi, i totally agree the discussion above. at least changes which impact install guide and could admin guide should have DocImpact.. 16:16:35 <jpich> david-lyle: More useful commit messages is good! 16:16:43 <jpich> -1 on a typo in the commit message is uncool in my world though :-) 16:17:26 <jpich> amotoki: Agreed. We probably should explore what's missing in these guides for Horizon and help the docs team fill the gaps as well 16:17:26 <david-lyle> yeah, let's not go crazy 16:17:27 <tzumainn> jpich, lol, agreed 16:17:37 <lsmola> jpich: agree, my vi doesn't have spellcheck 16:18:25 <amotoki_> Reagrding documentation, we use django style configuration and we need to explore how to sync our configurations with the config guide. 16:18:38 <jpich> lsmola: emacs has a spellchecker if you like ;) 16:18:49 <tzumainn> and jpich starts the flame war 16:19:02 <annegentle> do feel free to file doc bugs in http://bugs.launchpad.net/openstack-manuals for those missed DocImpact opportunities - you all will know better than us what should be docc'ed 16:19:18 <annegentle> amotoki_: it'd be great to automate those 16:19:24 <jpich> amotoki: Yep, which brings to mind https://bugs.launchpad.net/horizon/+bug/1300710 and https://bugs.launchpad.net/horizon/+bug/1221115 which we probably want in the guides as well 16:20:13 <jpich> annegentle: Thanks! That's what I had in mind, hopefully we can help with writing them too rather than just overload the docs team even more 16:20:47 <annegentle> jpich: sure! we are happy to take highly detailed doc bugs and turn them into docs if tools are an issue though, that's all I mean. 16:21:03 <jpich> annegentle: That's useful information, thanks! 16:22:03 <amotoki_> jpich: thanks for filing it. in other projects configurations are extracted from code base but it is not for Horizon. we need to explore how to improve it in Juno. 16:22:49 <david-lyle> yes, the horizon settings/conf mechanism is different and we'll have to provide a mechanism to pull it automagically 16:23:04 <jpich> yep 16:23:27 <david-lyle> anything else on this topic 16:23:37 <jpich> Not from me, thanks for your attention and consideration folks! 16:23:59 <amotoki_> nothing from me too. 16:24:28 <david-lyle> thanks! 16:24:36 <david-lyle> #topic Open Discussion 16:24:43 <tzumainn> so, related to my point above - is there a way to ask for patches to have more documentation associated with them, because a lot of those patches are fairly complex and it feels like a potential blocker for reviewers? 16:25:02 <david-lyle> tzumainn, absolutely 16:25:10 <annegentle> tzumainn: educate all your reviewers 16:25:44 <david-lyle> if you don't feel like enough information is provided to do a proper review, ask for it in the review 16:25:44 <tzumainn> so, I'm going to admit that by "reviewers" I meant "me" : ) 16:25:56 <jrist> so subtle 16:26:10 <jrist> what is everyone's thoughts on using polyfills or similar mechanisms for supporting older browsers with newer features (i.e. the recent 'input type number' bug) 16:26:13 <annegentle> tzumainn: I can't think of an automated way, or a test that runs each time 16:26:17 <jomara> tzumainn: your idea is going mainnstream 16:26:19 <david-lyle> you are likely not alone, and when looking back at the commit log, one sentence commit statements aren't very helpful 16:26:41 <jomara> tzumainn: ar eyou talking about commit docs, internal source comments, user guide stuff, or all of the above? 16:26:56 * david-lyle could have missed the point 16:26:57 <tzumainn> jomara, I'm talking about picking up a patch from gerrit and trying to figure out how to review it : ) 16:27:18 <jomara> so, whatever it is, enough to read about it and figure it out 16:27:25 <jpich> david-lyle: Agreed with all you said either way :-) 16:27:39 <jcoufal> jrist: I think we shouldn't pay so much extra effort to support old browsers, other than those which are officially supported 16:27:56 <jrist> do we have a list of officially supported browsers 16:28:04 <jrist> jcoufal: for instance, input type number is not supported in current firefox, 28 16:28:06 <jrist> pretty bad 16:28:13 <jrist> so not necessarily older browsers 16:28:30 <jcoufal> wait, really? current firefox doesn't support number type? 16:28:34 <jrist> nope 16:28:38 <jrist> 29 maybe. 28 no 16:28:38 * jcoufal amazed 16:28:56 <jrist> there are many things like that 16:29:29 <jcoufal> From my point of view, I would say that it depends on the type of the problem and browser 16:29:30 <david-lyle> the hard limitation we have is ie 8 and greater, other than that we assume the browsers do a good job of staying reasonably up-to-date 16:29:44 <david-lyle> soon ie 8 should go away 16:30:04 <jcoufal> general rule I have is - are you using latest new versions? you get great UX, are you using old stuff? it's usable for you, but you don't get the best you can 16:30:04 <jpich> There's some serious bugs open about our ie 8 support though I seem to remember 16:30:09 <jrist> is there a list somewhere 16:30:24 <jrist> of supported browsers 16:30:29 <jcoufal> i think david-lyle just described the rule above 16:30:43 <david-lyle> actually when XP eol'd we should be able to stop support ie8, it is also eol 16:30:51 <doug-fish> https://github.com/openstack/horizon/blob/master/doc/source/contributing.rst#javascript 16:30:55 <jomara> not bad! 16:31:04 <jcoufal> david-lyle: I can't wait for this to happen :) 16:31:13 <jpich> doug-fish: Thanks! I knew it was somewhere but my grepping failed me :) 16:31:26 <david-lyle> jcoufal, the date was April 8 16:31:27 <clu_> i think for older browsers, we should at least have a warning message somewhere that says: "horizon is not fully supported" 16:31:30 <david-lyle> it is done 16:31:32 <doug-fish> jpich: glad to help. Took me a while as well! 16:31:45 <tzumainn> so, there's a specific case here 16:31:49 <tzumainn> https://review.openstack.org/#/c/82908/ 16:31:50 <jpich> https://bugs.launchpad.net/horizon/+bug/1260281 - probably needs to be fixed and backported to icehouse 16:31:58 <tzumainn> the fix only works on these browsers: https://review.openstack.org/#/c/82908/ 16:32:04 <tzumainn> so the question is - should this fix be accepted or not? 16:32:05 <jrist> clu_: supported is a loaded word 16:32:10 <jcoufal> david-lyle: so we are safe to say, that for J we are dropping IE8 support as well 16:32:15 <jrist> tzumainn: thansk for bringing that up 16:32:24 <jrist> what do we mean by supported 16:32:30 <tzumainn> jrist, np, I had similar thought processes when looking at that patch 16:32:37 <david-lyle> jcoufal, seems so 16:32:40 <jrist> how about "Horizon and OpenStack Dashboard don't work well on browsers older than ...." 16:33:16 <tzumainn> jrist, so if that's the case, would the fix in the review above be accepted as closing the bug? 16:33:43 <david-lyle> jrist, more like "... are not supported on ... 16:33:57 <jrist> david-lyle: do we offer support? 16:34:08 <jrist> support is a loaded word, like I said 16:34:10 <david-lyle> jrist good point 16:34:22 <jrist> tzumainn: no I don't think it does, thus my initial nack 16:34:32 <amotoki_> Another idea is to have a list of tested browsers with versions. 16:34:33 * david-lyle thought jrist was the support organization 16:34:37 <jrist> I agree with Maxime 16:34:38 <tzumainn> jrist, yeah, I +1'd with an understanding it was a partial fix, but I can see your point 16:34:39 <jrist> david-lyle: ha 16:34:40 <david-lyle> :) 16:34:55 <clu_> amotoki_ - this is a good idea 16:35:25 <jrist> amotoki_: yeah, that is what I'm thinking 16:35:48 <jcoufal> jrist: I am on Maxime's side as well. It is not enough to use number type as a proper validation check 16:36:07 <jrist> so my original question 16:36:16 <jrist> what is everyone's thoughts on polyfills 16:36:23 <jrist> or similar mechanisms 16:36:32 <jrist> even IE9 doesn't support dozens of HTML5 things 16:37:00 <jcoufal> I am not very sure if the polyfill will ensure the validation 16:37:07 <jcoufal> isn't it only providing the look? 16:37:12 <jrist> it will, in the input number case 16:37:21 <jrist> in this case prevent non digit 16:37:58 <jcoufal> well then I would be in favor of that (in case of number picker) 16:38:10 <jrist> yeah, it is an example of other places where polyfills might be useful 16:38:26 <jcoufal> depends on case to case, but in this one it seems reasonable to me 16:38:34 <jrist> jcoufal: thanks. anyone else? :) 16:38:41 <jrist> I wish Maxime were here 16:38:58 <jpich> Can continue the discussion on channel later when he's around 16:39:03 <jrist> yeah 16:39:04 <jrist> ok thanks 16:39:55 <jpich> If it fixes things and the licence is compatible seems ok to me! 16:40:10 <clu_> i've seen some FIXME/TODOs in the code... do we do occasional sweeps to see if the issue can be resolved now? 16:40:42 <clu_> if so, write up bugs or something, otherwise things get forgotten and the FIXME/TODO list grows 16:41:19 <jpich> Ideally people would file a bug at the same time they submit a patch with a TODO 16:41:52 <doug-fish> I guess that's something we can add to our reviewer checklist 16:41:54 <david-lyle> clu_, that would be a good idea. Unfortunately, a lot of them are waiting on better support outside of Horizon, like server side filter criteria 16:42:25 <david-lyle> we could open a bug on something like that, but it would likely expire before we can address it 16:42:51 <clu_> david-lyle: agreed, we shouldn't write those up 16:43:01 <amotoki_> by greping the code, we have 77 TODOs but 30 TODOs comes from one person (router dashboard)..... Apart from this, we hvae ~40 TODO/FIXME. 16:43:03 <jpich> Ideally we'd add a horizon task to the bug in the related project but that's not always possible 16:43:19 <david-lyle> ideally :) 16:43:59 <jpich> It's nice to imagine the ideal world sometimes :-) 16:44:12 <david-lyle> amotoki_ 40 should be fairly easy to catagorize as actionable or not 16:44:20 <jrist> lol jpich 16:44:27 <tzumainn> I must say that it's nice that the cores in horizon are still optimists 16:44:30 <david-lyle> jpich, sometimes that's all we have 16:45:49 <amotoki_> can i discuss another topic? 16:46:02 <jpich> Please 16:46:19 <amotoki_> i would like to know opinions on backward compat of "horizon" module. 16:46:31 <amotoki_> https://review.openstack.org/#/c/86068/2/horizon/tables/actions.py 16:46:47 <amotoki_> this patch removes existing fields in BacthAction. 16:47:08 <amotoki_> I wonder we should consider backward-compat of horizon module or not. 16:47:29 <jpich> We will have to be better at it when we separate it from the dashboard for sure 16:48:29 <amotoki_> fair enough to me. 16:48:46 <jpich> This might be a topic worthy of a more meaty/thoughtful conversation on list... 16:49:39 <amotoki_> agree. i just want to know initial feedbacks. 16:50:19 <david-lyle> I think existing fields should be a high hurdle 16:50:52 <david-lyle> as an extensible framework, we potentially break existing deploys 16:51:50 <amotoki_> yes. that is in my mind. I remember Gabriel mentioned so in past reviews. 16:52:17 <david-lyle> I have to look at this particular patch in more depth to understand the intent better 16:53:08 <david-lyle> I think the story will actually be smoother after the split, because we will have packaged versions of the toolkit that doesn't have to be updated with Horizon 16:53:35 <david-lyle> makes our testing matrix more complex of course 16:54:03 <david-lyle> s/Horizon/openstack_dashboard_future_name/ 16:55:10 <santibaldassin> hey folks..I'd like to have a quick talk about client side validations if there's no other topic 16:55:19 <david-lyle> sure 16:56:00 <santibaldassin> so, there are a couple of patches were we are doing an api call to get quota usages do do client side validations that are already done by nova 16:56:12 <santibaldassin> https://review.openstack.org/#/c/76107/ 16:56:19 <santibaldassin> https://review.openstack.org/#/c/70158/14/openstack_dashboard/dashboards/admin/projects/workflows.py 16:58:44 <jpich> It's usually nicer to show an error on the form by the field that has the issue, rather than let the user submit it then have to start from scratch again to correct an error 16:58:52 <santibaldassin> I have two concerns, one is that we are increasing the number of request that we are doing to complete the operation and the other is that if the backend *no matter if its nova or cinder or any other module) changes the way they do the validation, we'll have to change the validation in horizon too 17:00:01 <jpich> Although for that one in particular, do the APIs actually prevent you from setting a quota less than what is currently used? 17:00:34 <david-lyle> let's continue this in the horizon room or on the review 17:00:34 <jpich> well, we're out of time, we can continue on the channel - although I have to jump in another meeting now 17:00:35 <santibaldassin> jpich: I'm ok with showing friendly user. I think we could do it without the extra api calls 17:00:50 <david-lyle> Thanks everyone have a great week. 17:00:53 <santibaldassin> thanks 17:00:56 <david-lyle> #endmeeting