13:00:16 <alex_xu> #startmeeting nova api
13:00:16 <openstack> Meeting started Wed Aug 24 13:00:16 2016 UTC and is due to finish in 60 minutes.  The chair is alex_xu. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:17 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:19 <openstack> The meeting name has been set to 'nova_api'
13:00:25 <alex_xu> who is here today?
13:00:26 <johnthetubaguy> o/
13:00:29 <jichen> o/
13:00:57 <alex_xu> Sean is not here this week due to some openstack event in the US
13:01:19 <alex_xu> so let us start the meeting
13:01:26 <alex_xu> #topic API Priorities
13:01:42 <alex_xu> as I know, we just left two patches for newton
13:02:21 <alex_xu> microversion 2.38 merged. deprecate network CLI merged
13:02:31 <johnthetubaguy> (until the api-ref push, I guess)
13:02:52 <alex_xu> johnthetubaguy: ah, thanks, you are right
13:03:10 <johnthetubaguy> so there is a policy one from alaski
13:03:26 <alex_xu> s/left two patches for newton/left two patches before freeze/ :)
13:03:33 <alex_xu> #link https://review.openstack.org/356963
13:03:43 * edleafe wanders in late
13:04:01 <alex_xu> johnthetubaguy: have you checked Sean's comment?
13:04:17 <johnthetubaguy> not seen it
13:05:17 <johnthetubaguy> #link https://review.openstack.org/#/c/335667/
13:05:17 <alex_xu> johnthetubaguy: so just some check without any IO, so is the performance problem ok?
13:05:24 <johnthetubaguy> policy one^
13:06:01 <alex_xu> johnthetubaguy: thanks, another one I missed
13:06:42 <alex_xu> #link https://review.openstack.org/322944
13:06:48 <alex_xu> policy cli ^
13:08:06 <johnthetubaguy> so we should go through those I guess
13:08:20 <johnthetubaguy> #link https://review.openstack.org/#/c/356963/4/nova/policy.py
13:08:28 <johnthetubaguy> what way do we feel we should go with this one
13:08:52 <johnthetubaguy> does it look terrible to not do that check every time?
13:10:31 <alex_xu> johnthetubaguy: yes, I feel user like change that in runtime?
13:12:28 <johnthetubaguy> do we know how long that operation takes?
13:12:42 <johnthetubaguy> I guess it might be quick
13:13:06 <alex_xu> johnthetubaguy: yes, no IO happened in that check
13:13:19 <alex_xu> I have no exact number
13:14:18 <johnthetubaguy> I just wonder if we should complie the regular expression, I can't remember if that is still needed these days
13:14:38 <johnthetubaguy> just checked, its not required
13:14:53 <johnthetubaguy> assuming we are a program that only uses a few regular expressions at a time
13:15:18 <alex_xu> I can check that or just compile it
13:15:20 <edleafe> johnthetubaguy: yeah, that's usually only helpful for things used very often
13:15:36 <johnthetubaguy> edleafe: you mean like something that gets used on every API call?
13:15:57 <edleafe> johnthetubaguy: well, yeah, that's usually a good use case :)
13:17:06 <johnthetubaguy> alex_xu: so I should just go check that, and see how slow it really is
13:17:29 <alex_xu> johnthetubaguy: ok, thanks, I will check that also.
13:18:42 <alex_xu> so, any more question for this patch?
13:19:47 <alex_xu> I guess no
13:20:10 <alex_xu> #link https://review.openstack.org/322944
13:20:26 <alex_xu> johnthetubaguy: I think I should write a release for this patch?
13:20:43 <alex_xu> /release/release note/
13:21:14 <johnthetubaguy> alex_xu: hmm, we probably should, its a new feature
13:21:31 <alex_xu> johnthetubaguy: ok, got it, I will update that
13:21:47 <johnthetubaguy> I noticed it calls init() in policy, that isn't required
13:23:59 <alex_xu> johnthetubaguy: we need to filter the rules at https://review.openstack.org/#/c/322944/10/nova/cmd/policy_check.py@133
13:24:45 <johnthetubaguy> anyways, maybe what I am getting is we should all go away and review those three patches carefully, and just end the meeting now?
13:25:13 <alex_xu> johnthetubaguy: ah, you are right, this cmd only enforce on the default rule
13:25:25 <alex_xu> johnthetubaguy: yea, agree
13:26:15 <alex_xu> so any other question want to bring up? otherwise let us back to work.
13:26:28 <johnthetubaguy> alex_xu: we could call it just before list_rules, that could make some sense
13:26:40 <johnthetubaguy> but actually, yeah, that doesn't change
13:28:11 <alex_xu> johnthetubaguy: yea
13:28:15 <johnthetubaguy> this cmd, I wonder if we really want to return Pass/Fail for each rule we have checked
13:28:55 <alex_xu> johnthetubaguy: you mean the full-list?
13:29:04 <johnthetubaguy> no, just the matched ones
13:29:33 <alex_xu> johnthetubaguy: that is the current patch doing
13:29:52 <johnthetubaguy> I was unclear
13:30:11 <johnthetubaguy> I was thinking we could return the ones we pass or fail, but I remember why we don't do that now
13:30:29 <johnthetubaguy> i.e. matches the api-name
13:30:42 <johnthetubaguy> but the context.can result is just reported
13:31:05 <johnthetubaguy> foo:a  PASS  foo:b FAIL   foo:c PASS, etc
13:31:55 <johnthetubaguy> anyways, ignore me
13:31:58 <johnthetubaguy> lets just go review things
13:32:14 <alex_xu> johnthetubaguy: I kind of like your idea, good for user to know how much rules we have
13:32:51 <alex_xu> johnthetubaguy: ok
13:33:01 <johnthetubaguy> I think the cmd was going to try match the API I guess
13:33:07 <johnthetubaguy> hence the different approach
13:33:44 <alex_xu> johnthetubaguy: the parameter '--api-name' is really a wrong name for current implementation
13:34:15 <alex_xu> but looks like no really easy way to match the rule and api
13:34:51 <alex_xu> anyway let us back to work, or just discuss the patch off the meeting
13:35:00 <alex_xu> so thanks all!
13:35:05 <alex_xu> #endmeeting