16:00:11 <NikolayM> #startmeeting Mistral
16:00:12 <openstack> Meeting started Mon Feb 16 16:00:11 2015 UTC and is due to finish in 60 minutes.  The chair is NikolayM. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:15 <openstack> The meeting name has been set to 'mistral'
16:00:21 <NikolayM> hi everyone!
16:01:16 <NikolayM> anyone else here?
16:01:58 <dzimine> hi.
16:02:25 <NikolayM> hi dzimine
16:02:36 <akuznetsova> hey
16:02:49 <NikolayM> ok, let's start
16:02:59 <akuznetsova> Renat is absent today
16:03:06 <NikolayM> yes
16:03:12 <NikolayM> #topic Review Action Items
16:03:26 <NikolayM> 1. rakhmerov, dzimine: capture everything in blueprints
16:03:52 <dzimine> not yet done on my site :(
16:04:00 <dzimine> will complete this week.
16:04:08 <NikolayM> ok
16:04:19 <NikolayM> I just leave new AI :)
16:04:41 <NikolayM> #action rakhmerov, dzimine: capture everything in blueprints
16:04:55 <NikolayM> 2. rakhmerov, dzimine: review existing bugs and assign them to the scope
16:05:11 <NikolayM> It seems it's done
16:05:34 <NikolayM> 3. rakhmerov: talk to solum folks (Roshan) and find out what version of Mistral they depend on
16:05:51 <NikolayM> ooh, I forgot to ask Renat about it
16:06:03 <dzimine> he's not around, I gather?
16:06:10 <akuznetsova> Renat didn't talk about it on our sync-ups
16:06:40 <NikolayM> #action rakhmerov: talk to solum folks (Roshan) and find out what version of Mistral they depend on
16:06:54 <NikolayM> yes, we didn't hear from him about it
16:06:56 <akuznetsova> probably he forgot about it
16:07:19 <NikolayM> 4. all: think if we need to keep v1 code base
16:07:56 <NikolayM> I guess StackStorm guys ready to drop v1
16:08:30 <NikolayM> And since we don't know how about Solum, leave this action item on next meeting.
16:08:53 <NikolayM> #action all: think if we need to keep v1 code base
16:09:22 <NikolayM> #topic Current status
16:10:00 <NikolayM> let's tell about our progress
16:10:08 <NikolayM> my status:
16:10:26 <NikolayM> worked on YAQL expressions and its escaping
16:11:25 <akuznetsova> I am working on OpenStack actions simple integration tests
16:11:53 <NikolayM> and did some research about keystone/trusts, how we can do heat/mistral integration. So, created new thread in ML: "How trusts should work by design?"
16:15:02 <NikolayM> the next topic is kilo-3 planning
16:15:26 <NikolayM> #topic Kilo-3 planning
16:16:34 <NikolayM> We need to redesign our engine
16:17:07 <NikolayM> it is a big work and Renat already started on it
16:17:34 <dzimine> sorry I had to step out.
16:17:58 <dzimine> back now.
16:18:12 <NikolayM> We think about how to share this work between us
16:18:26 <dzimine> the engine redesign? yes.
16:18:32 <NikolayM> yes
16:18:47 <NikolayM> And it seems it will take a lot of time to do it
16:19:16 <akuznetsova> NikolayM, we want to add 'action handler', am I correct ?
16:19:28 <NikolayM> so, we can't do some too serious things to kilo-3
16:20:20 <NikolayM> no, AFAIU we want to add something like on_action_result
16:21:01 <dzimine> On my side: we got strong negative feedback from the field on our changes to YAQL syntax. The crux of the problem is quoting expressions the ansible way like "{1+1}" is confusing, but not quoting it is impossible as {} are core YAQL syntax symbols (dictionary).
16:21:13 <NikolayM> so, we will add ActionExecution model and Task will be compound thing
16:21:41 <NikolayM> dzimine, yes
16:21:53 <NikolayM> let's discuss this
16:21:57 <akuznetsova> dzimine, we are thinking about other syntax
16:22:05 <dzimine> I spoke to Renat on this and we realized that we can't continue using YAML syntax for expressions no matter what others (salt, ansible) do.
16:22:12 <NikolayM> I know we want to change YAQL syntax
16:22:41 <dzimine> The proposal on the table now is to change {1+1} to <% 1 + 1 %> and it solves a bunch of problems.
16:22:47 <NikolayM> And I heard you suggested the syntax like <% %>
16:22:53 <dzimine> I am preparing email to the thread.
16:23:27 <akuznetsova> dzimine, why did you choose such syntax ?
16:23:30 <dzimine> The reason for this particular syntax is our target user's familiarity: many of them do Puppet and Chef and it's common Ruby pattern one sees in Puppet and Chef.
16:24:21 <NikolayM> <% %> construction look not good, what about discussion on other syntax?
16:24:26 <dzimine> the critera for the choice is "plays well in YAML" and "target users are familiar / seen that before "
16:24:50 <akuznetsova> as for me it looks familiar with xml/html
16:25:12 <dzimine> old JSP also used this. and Ruby.
16:25:33 <NikolayM> 2 symbols is cool but percent sign is too big for perception
16:26:06 <dzimine> too big for perception - what do you mean?
16:26:25 <NikolayM> percent sign is just too big
16:26:45 <NikolayM> and overall structure become x2 bigger
16:27:42 <dzimine> visually?
16:27:42 <akuznetsova> I like something like <yaql 1 + 1 >, I know that it is too big, but as for me it is clear for undestanding
16:27:50 <NikolayM> dzimine, yes
16:28:31 <dzimine> we considered HEAT/TOSCA approach for this using functions: similar to what akuznetsova mentioned ^^
16:29:30 <dzimine> but it seem to get messy when doing complex yaql in combinations with string building and other parameters.
16:29:39 <NikolayM> looks better than with % but too long
16:29:48 <dzimine> the two symbols solve the problem of escaping.
16:30:30 <dzimine> Here is the thing: when you or I look at the problems, we look at them differently when the guys whom we want to use this thing.
16:31:16 <dzimine> So it's interesting to listen field folks on what they like/not like.
16:31:31 <dzimine> That's why most of Mirantis products are pretty good: they grew out of the field use.
16:31:47 <dzimine> Mistral - not yet, we just began deploying it to customers.
16:31:58 <dzimine> And their complains are pretty .... unexpected.
16:32:30 <dzimine> I was arguing for Ansible syntax for over an hour, wanted to keep "{1+1}"
16:33:10 <dzimine> but they explained to me why it was confusing to them, on examples.
16:33:11 <NikolayM> I also agree we don't use {} due to YAML constraints
16:33:37 <dzimine> the way to come to agreement is to arrive at criteria, and rank the criteria.
16:33:49 <dzimine> That's why I think criteria #1 is "YAML friendly".
16:33:55 <dzimine> #2 is "familiarity".
16:34:44 <NikolayM> yes
16:34:45 <dzimine> and there's more in this "familiarity": a number of engines already use {{ or <% so they figured that it's safe to parse / escape / etc.
16:35:07 <NikolayM> We need to think about it carefully
16:35:36 <dzimine> it's ok if we invent somethign new, but say if we only do <  and > than we need to begin worrying about the other meanings of the symbols, escaping the greater than and less than signs.
16:36:06 <dzimine> But that the guys still find it confusing: say "I don't know the meaning of < > when I see them in a large expression.
16:36:38 <dzimine> We discussed just < and > but all these questions came up.
16:37:31 <dzimine> By now I think <% is a good syntax: not visually as nice as NikolayM pointed out but quite distinct and once we get used to it will be ok.
16:38:19 <dzimine> do bring alternatives but please do find something familiar.
16:39:16 <dzimine> And BTW we need this change ASAP: our recent changes with turning YAQL with "{1+1}" are breaking existing workflows.
16:39:22 <dzimine> so we don't want to break once, than again.
16:40:01 <NikolayM> ok
16:40:41 <NikolayM> dzimine, let's take a meeting with Renat and discuss all variants
16:40:55 <dzimine> ok.
16:41:08 <dzimine> I'll start a thread on openstack-dev, too.
16:41:35 <NikolayM> ok, I also will take part in
16:42:23 <dzimine> anything else to discuss?
16:42:30 <akuznetsova> we can do <? ..?> like php injection)
16:42:41 <dzimine> yes we can I think.
16:42:59 <dzimine> but again, when you look at your target users, assume they do php?
16:44:25 <akuznetsova> no
16:46:03 <NikolayM> well, if we have no more to discuss, let's wrap up
16:46:49 <NikolayM> we'll discuss our new YAQL syntax in ML and (I think) in skype/hangouts with Renat
16:46:58 <dzimine> ok.
16:47:05 <dzimine> thanks all! Till than!
16:47:16 <NikolayM> bye dzimine!
16:47:22 <akuznetsova> bye all
16:47:48 <NikolayM> #endmeeting