00:02:32 <cyeoh> #startmeeting nova api wg
00:02:33 <openstack> Meeting started Thu Mar 19 00:02:32 2015 UTC and is due to finish in 60 minutes.  The chair is cyeoh. Information about MeetBot at http://wiki.debian.org/MeetBot.
00:02:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
00:02:36 <openstack> The meeting name has been set to 'nova_api_wg'
00:02:43 <elmiko> so far i think it's just you and i
00:02:55 <cyeoh> ah ok big meeting then :-)
00:03:30 <cyeoh> since there is just the two of us on what is the most important thing on your mind?
00:04:23 <elmiko> hmm
00:04:38 <elmiko> nothing big api-wise for me this week
00:04:56 <elmiko> biggest issue for sahara team atm is the api-ref wadl stuff
00:05:02 <elmiko> but i think we've got it under control
00:05:17 <elmiko> how about you?
00:06:07 <cyeoh> standardisation-wise probably just project-client issues because I'm trying to get a new novaclient out that we can use microversions
00:06:42 <elmiko> do you have a review to link?
00:06:42 <cyeoh> but most of it at this stage is primarily making it up as i go along as we'll see how it works - since its all a bit new
00:06:51 <elmiko> ah, ok
00:07:04 <cyeoh> https://review.openstack.org/#/c/152569/17
00:07:30 * elmiko looks
00:08:05 <cyeoh> that one is andrew's. Im planning on splitting it up into smaller changesets to make it easier to review (eg - some for cleanup first etc) and then a series of small patches that only add a bit of functionality at a time
00:08:28 <elmiko> nice
00:08:38 <elmiko> the microversion stuff is represented all throught header?
00:08:40 <cyeoh> eg since we have 2 API implementations and theoretically 4 apis need to be able to select them from novaclient
00:09:17 <cyeoh> yes, so its all gets boiled down to one header. Whether its there or not and then what its value is
00:09:35 <elmiko> and the major version part is represented in the uri?
00:10:50 <cyeoh> currently it does, but in the future we will be moving away from that and it will be solely header dependent
00:11:01 <cyeoh> https://etherpad.openstack.org/p/novaclient_microversions_design
00:11:11 <elmiko> ah, thanks
00:11:15 <cyeoh> this is a bit of a high level look I've been working on
00:12:09 <elmiko> i don't have a strong opinion on the version in header vs in uri, but it seemed to draw out a good conversation
00:12:19 <cyeoh> we do want to move url discovery out of keystone (eg no more compute and computev3 etc
00:12:27 <elmiko> nice
00:13:38 <cyeoh> yea there are lots of opinions out there - but with microversions and lots of version bumping putting versions in the url path would be a lot of overhead for clients eventually
00:13:56 <elmiko> yea
00:15:05 <cyeoh> so is the api-ref wadl for sahara intended to be automated?
00:15:26 <elmiko> unfortunately no
00:15:29 <cyeoh> background goal for us in Nova is automation of api ref docs.
00:15:32 <cyeoh> ah ok.
00:15:54 <elmiko> i'd like to work towards automation as well, but i don't think wadl is the format for it
00:16:13 <elmiko> i've been investigating swagger a bunch with our codebase and the barbican code
00:16:39 <elmiko> the wadl in api-ref seems like it needs a bunch of hand tweaking to really get it thorough
00:17:19 <elmiko> i think a big issue with automation across the board will be projects that use pecan
00:17:57 <cyeoh> yea and soon as manual intervention is required we get errors.
00:18:07 <elmiko> right, plus the maintenance costs
00:18:22 <elmiko> i think a good step would be partial automation to start with
00:18:33 <elmiko> but i think the doc team doesn't want full automation, with good reason
00:19:29 <cyeoh> yea I'm happy with that. I just want things like paramter names/types automated. Still found those sorts of errors just 2-3 months ago
00:19:50 <elmiko> agreed
00:20:14 <elmiko> it'll be interesting to see where this goes
00:20:34 <elmiko> i feel like the swagger json format is much more friendly to interact with than the wadl xml
00:20:45 <elmiko> so you could generate a skeleton, and fill in the details
00:20:59 <cyeoh> Boo to XML!
00:21:03 <elmiko> hehe
00:22:17 <cyeoh> yep sounds like we could still reverify ocassionally too so an updated skeleton still fits what we publish
00:22:45 <elmiko> right
00:22:52 <cyeoh> and we don't have doco for features which actually don't exist anymore :-)
00:23:03 <elmiko> exactly
00:24:46 <elmiko> i have a feeling next meeting we will discuss meeting times ;)
00:25:08 <cyeoh> meeting times are hard :-(
00:25:12 <elmiko> hehe
00:25:34 <cyeoh> this and the nova api meeting is the only one I have that is NOT between 12am and 6am
00:25:47 <elmiko> ouch
00:25:59 <cyeoh> and I really can't do those sorts of really early times for a while
00:26:01 <elmiko> thankfully i don't have any in that range
00:27:29 <cyeoh> yea I'm fine with the other meeting week moving to a beter US/EU time, but I'd like to keep something that works for Aust/NZ/Asia
00:27:42 <elmiko> +1
00:28:13 <elmiko> i'm in the US, where are you at?
00:28:40 <cyeoh> I'm in Adelaide Australia (GMT+10:30)
00:28:48 <elmiko> cool
00:29:05 <cyeoh> so this time actually works really well ;-)
00:29:08 <elmiko> yea
00:29:15 <elmiko> i'm GMT-4 currently, i think
00:29:27 <elmiko> yea -4
00:29:48 <cyeoh> heh my meetings with US people are generally at interesting times for at least one of us
00:29:55 <elmiko> lol, i'll bet
00:30:20 <elmiko> i wonder how many folks we have from Aust/NZ/Asia who participate in the api-wg
00:31:06 <cyeoh> a couple from Japan at least, and then at least one from China (but that will probably grow)
00:31:48 <elmiko> nice
00:32:19 <elmiko> the sahara team has had a recent influx of folks from asia, it's nice to have more hands onboard =)
00:33:52 <cyeoh> yea definitely! Sorry just got pinged by my manager so am going to have to go. Will be back later
00:34:07 <elmiko> cyeoh: end the meeting before you go
00:35:00 <cyeoh> #endmeeting