00:01:18 <ekcs> #startmeeting congressteammeeting
00:01:19 <openstack> Meeting started Thu Oct 12 00:01:18 2017 UTC and is due to finish in 60 minutes.  The chair is ekcs. Information about MeetBot at http://wiki.debian.org/MeetBot.
00:01:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
00:01:22 <openstack> The meeting name has been set to 'congressteammeeting'
00:01:27 <thinrichs> Hi all
00:01:30 <ekcs> hello all.
00:01:32 <ekcs> hi thinrichs
00:01:46 <ekcs> topics here as usual: https://etherpad.openstack.org/p/congress-meeting-topics
00:03:00 <masahito> ekcs: hi.
00:03:09 <ekcs> hi masahito !
00:03:22 <ekcs> topics here as usual: https://etherpad.openstack.org/p/congress-meeting-topics
00:03:40 <masahito> sorry, I need to go in 30 mins.
00:03:55 <ekcs> got it!
00:04:16 <ekcs> well let’s get started then.
00:04:19 <ekcs> #topic priorities
00:04:52 <ekcs> since the PTG I’ve been maintaining a list of congress priorities in strict order.
00:04:52 <ekcs> https://etherpad.openstack.org/p/congress-task-priority
00:05:15 <ekcs> Idea is to avoid having a bunch of things all at the same priority level.
00:05:41 <ekcs> in reality it may not be exact, but it’s better than having 10 “high priority” items we can’t distinguish between.
00:06:18 <ekcs> Please take a look from time to time and make any comments, suggestions, changes as needed.
00:06:19 <thinrichs> Total priority is good
00:06:52 <ekcs> yups.
00:07:31 <ekcs> we’re making decent progress so far. 1, 2.2, 2.3, 4 are all in review.
00:07:56 <ekcs> 1 is ready to merge just need the unblock the gate still.
00:08:46 <ekcs> any comments, additions, thoughts, or discussions on the list?
00:10:32 <thinrichs> Look good to me.  I'd focus on solving community pain points, which it looks like is where the priorities are lining up.
00:11:04 <ekcs> thinrichs: agreed on focusing on concrete pain points.
00:11:40 <ekcs> please feel free to insert anything you guys see missing from the list.
00:11:48 <ekcs> or bring them up in IRC.
00:14:27 <ekcs> ok moving on then
00:15:08 <ekcs> #topic gate blockage
00:15:30 <ekcs> unfortunately our gate is still blocked since the jenkin to zuul migration.
00:16:19 <ekcs> we have patch ready to fix things in our repo, but that patch is still blocked by stuff in other repos.
00:16:43 <ekcs> also seeing a fair amount of infra instability with POST_FAILURE.
00:17:01 <ekcs> hoping for things to be settled down by next week.
00:17:22 <ekcs> quite a few patches stuck at the gate.
00:17:33 <ekcs> any thoughts/comments/questions?
00:18:05 <ekcs> main external issue we’re still waiting on: https://bugs.launchpad.net/designate/+bug/1721432
00:18:06 <openstack> Launchpad bug 1721432 in oslo.context "TypeError: __init__() got an unexpected keyword argument 'project' in Designate" [Critical,In progress] - Assigned to Juan Antonio Osorio Robles (juan-osorio-robles)
00:19:00 <ekcs> ok moving along then.
00:19:16 <ekcs> #topic continue drivers conf discussion
00:20:14 <ekcs> actually ramineni seems to be in and out.
00:20:23 <ekcs> maybe we’ll wait a bit more.
00:20:31 <ekcs> any other topics on our minds?
00:20:56 <thinrichs> Not me
00:21:04 <thinrichs> except… did you figure out the grammar problem?
00:21:53 <ekcs> thinrichs: no I haven’t. the triple quote appears to be broken. but not sure we want to actually try and fix it.
00:22:36 <ekcs> I’m thinking we’ll work around it for feeding YAML to mistral in rule.
00:23:16 <ekcs> like explicitly de-escape the \n in a string argument before passing on the mistral.
00:24:03 <ekcs> one issue is what the intended result is when we write something like p(“a\nb”) in a congress rule.
00:24:35 <ekcs> is it supposed to be length 3 with a newline character or is it supposed to be length 4? right now it reads as length 4.
00:25:12 <ekcs> I’m thinking we don’t really want to muck with that behavior so we can just add another layer of de-escape on top of that.
00:25:21 <ekcs> open to thoughts and suggestions.
00:27:14 <ekcs> btw here is what I organized from our previous discussion around changing driver conf behavior: https://etherpad.openstack.org/p/congress-driver-conf
00:28:07 <thinrichs> Not clear to me why embedding the carriage-return in the triple quote is the wrong thing.  It gives the user the power to choose that.
00:28:16 <thinrichs> Especially if the triple-quotes are broken.
00:28:43 <thinrichs> I also don't know where the code that would change \n to a CRLF would live.  The datasource?
00:29:21 <ekcs> thinrichs: I’m thinking the datasource, for specific use like YAML string arg.
00:29:43 <ekcs> thinrichs: I agree it’s cleaner to make p(“a\nb”) read as length 4 string.
00:29:49 <ekcs> sorry length 3
00:30:39 <thinrichs> To be clear, I was thinking "a\nb" would be a 4-length string, but """a\nb""" would be a 3-length string.
00:30:57 <ekcs> oh I see what you mean.
00:31:33 <ekcs> I was thinking “a\nb” would be length 3, and “a\\nb” would be length 4. consistent with python.
00:31:50 <ekcs> so you’re saying since triple quote is broken, we can just fix it.
00:32:01 <ekcs> since clearly we’re not going to break people.
00:32:32 <ekcs> in python triple quote, “””a\nb””” would be length 4.
00:32:37 <ekcs> “””a
00:32:40 <ekcs> b”””
00:32:45 <ekcs> would be length 3.
00:33:06 <thinrichs> Oh I see.  I was assuming we wanted to write with the actual line-breaks in there.
00:33:32 <thinrichs> But you're saying we want to use \n instead of the actual line-break.
00:34:15 <masahito> sorry to interrupt. I need to go.
00:34:17 <thinrichs> If I were embedding YAML into the file I'd probably want to cut-and-paste as is (and use YAML linters).  That would require using the actual line breaks, instead of the \n./
00:34:35 <ekcs> thanks masahito see you later!
00:34:42 <masahito> ekcs: thanks, bye.
00:35:09 <thinrichs> bye masahito
00:35:16 <masahito> thinrichs: bye
00:35:36 <ekcs> thinrichs: right. if we want to be able to copy past YAML into an arg, we’d need to fix triple quote and allow actual line breaks in the input.
00:35:55 <ekcs> that would be ideal.
00:36:29 <ekcs> but I have no idea where to start in fixing triple quote.
00:37:00 <ekcs> and a compromise would be for user to replace newline by \n before pasting.
00:37:28 <ekcs> may be the preferable style anyway in many cases.
00:37:45 <ekcs> rather than
00:38:22 <thinrichs> So then something consistent with Python makes sense.  Maybe a backwards incompatible change, but that seems fine.
00:40:01 <ekcs> what is the backward incompatible change?
00:40:14 <ekcs> reading “a\nb” is length three?
00:40:54 <thinrichs> Yes
00:41:00 <thinrichs> That's not how it works today, right?
00:41:38 <ekcs> no not how it works today. my favored option right now is to keep it working the way it is today.
00:41:44 <thinrichs> (It'd obviously be a pretty tiny change in behavior)
00:41:59 <thinrichs> Oh—then no backwards incompatible change.
00:42:11 <thinrichs> Put the translation in the datasource driver.
00:42:14 <thinrichs> Right?
00:42:18 <ekcs> but simply built it into specific action executions to de-escape. so it’s part of that action semantic rather than general parsing semantic
00:42:36 <ekcs> right. it’s a bit hacky (ie. specific to something)
00:42:51 <ekcs> but it keeps backward compatibility.
00:45:28 <ekcs> but it’s also an option to just change the way “a\nb” parses completely to be consistent with python.
00:46:17 <ekcs> i’m not sure how to do that either, but it may be easier to figure out than the triple quote.
00:47:42 <thinrichs> The datasource answer sounds like the simplest for now
00:48:26 <ekcs> yea that’s what I’m thinking.
00:48:35 <ekcs> i’ll give it a try.
00:48:49 <ekcs> so do you have any more thoughts on the drivers conf?
00:49:42 <ekcs> I added a new suggestion which is to treat the existing drivers option as specifying drivers in addition to the default ones.
00:49:50 <ekcs> https://etherpad.openstack.org/p/congress-driver-conf
00:50:30 <ekcs> that way people who have additional drivers configured don’t get broken by upgrade. people who have old config don’t miss out on new drivers.
00:50:57 <ekcs> only people who get “messed up” are people who specifically want certain drivers to be unavailable.
00:51:19 <ekcs> they’d need to specify a new whitelist_driver option to limit to those.
00:51:54 <ekcs> that’s 2.2 in the etherpad.
00:53:03 <ekcs> sorry it’s 2.3 in the etherpad.
00:53:13 <ekcs> 2.2 is what I think both you and ramineni favored in last meeting.
00:53:47 <ekcs> 2.1 is what I favored last meeting before I considered 2.3.
01:00:03 <ekcs> ok well time’s up. thanks for the discussion! feel free to add comments to the etherpad and/or respond to my ML message later. have a great week. #endmeeting
01:00:09 <ekcs> #endmeeting