| *** amitgandhinz has quit IRC | 00:01 | |
| *** amitgandhinz has joined #openstack-zaqar | 00:09 | |
| *** amitgandhinz has quit IRC | 00:22 | |
| *** jeffrey4l has joined #openstack-zaqar | 00:36 | |
| wpf | jeffrey4l: morning | 01:08 |
|---|---|---|
| jeffrey4l | wpf, morning sir! | 01:08 |
| wpf | some questions about your PS, please check the review , -:) | 01:09 |
| * jeffrey4l is checking. | 01:10 | |
| *** amitgandhinz has joined #openstack-zaqar | 01:12 | |
| jeffrey4l | wpf, pls check my comments for your questions. | 01:21 |
| flwang | wpf: pengfei, do you know who from IBM will join the summit? | 01:23 |
| wpf | jeffrey4l: xuhan, haiming , I thought | 01:28 |
| wpf | do you know them? | 01:28 |
| jeffrey4l | wpf, u ping to wrong person.. | 01:28 |
| wpf | -:) | 01:28 |
| wpf | flwang: xuhan, haiming , I thought | 01:29 |
| flwang | wpf: ok, cool, thanks | 01:29 |
| wpf | flwang: you also will be there ,right? | 01:29 |
| flwang | wpf: what's your focus now? | 01:29 |
| flwang | wpf: yep, hope it won't be cancelled again :D | 01:29 |
| wpf | still hybrid , but not too much amazing features .............. | 01:30 |
| wpf | :P | 01:30 |
| flwang | wow | 01:31 |
| *** bradjones has quit IRC | 01:50 | |
| *** bradjones has joined #openstack-zaqar | 01:50 | |
| openstackgerrit | Jeffrey Zhang proposed a change to openstack/zaqar: Always include the project id in the logs https://review.openstack.org/130139 | 02:05 |
| *** amitgandhinz has quit IRC | 03:06 | |
| *** exploreshaifali has joined #openstack-zaqar | 03:45 | |
| *** exploreshaifali has quit IRC | 04:23 | |
| *** flwang1 has quit IRC | 04:26 | |
| *** gmann has joined #openstack-zaqar | 04:33 | |
| gmann | flaper87: hi | 04:35 |
| gmann | flaper87, flwang: 1 quick question - is delete queue operation is sync or async? | 04:42 |
| gmann | flaper87, flwang : Actually I was having a look on tempest tests where it is tested as sync. | 04:43 |
| *** flwang1 has joined #openstack-zaqar | 05:06 | |
| *** sgotliv has joined #openstack-zaqar | 05:08 | |
| jeffrey4l | gmann, it is sync. Why u have such a question? | 05:28 |
| *** exploreshaifali has joined #openstack-zaqar | 05:33 | |
| gmann | jeffrey4l: thx, actually tempest tests that operation as sync, so i wanted to confirm the same. | 05:33 |
| jeffrey4l | got it | 05:34 |
| *** flwang1 has quit IRC | 05:47 | |
| *** yfujioka has joined #openstack-zaqar | 05:57 | |
| *** openstack has joined #openstack-zaqar | 06:12 | |
| *** exploreshaifali has quit IRC | 06:15 | |
| *** prashanthr_ has joined #openstack-zaqar | 06:15 | |
| *** jeffrey4l has quit IRC | 06:40 | |
| *** openstack has quit IRC | 06:59 | |
| *** openstack has joined #openstack-zaqar | 14:18 | |
| *** openstackstatus has joined #openstack-zaqar | 14:19 | |
| *** ChanServ sets mode: +v openstackstatus | 14:19 | |
| *** mpanetta_ is now known as mpanetta | 14:19 | |
| flaper87 | vkmc: right, but as it is, it just has support for v1.0 and not v1.1 (I'm talking about the case where just 1 message is posted) | 14:43 |
| flaper87 | vkmc: does that make sense? | 14:44 |
| flaper87 | vkmc: for example: queue.post(dict(field1='test')) | 14:44 |
| flaper87 | That fails for v1.1 | 14:45 |
| flaper87 | because it won't wrap it in {'messages': [dict(field1='test')]} | 14:45 |
| vkmc | yeah, but that is on the client | 14:45 |
| vkmc | we assume that if there is no 'messages' in the dict then we are talking about v1 | 14:46 |
| vkmc | I could check the client version and stop assuming | 14:46 |
| vkmc | and... it works for v1.1, when just 1 message is posted :) | 14:47 |
| flaper87 | mmh, does it? | 14:47 |
| flaper87 | I | 14:48 |
| flaper87 | I must be missing something then | 14:48 |
| flaper87 | let me check that again | 14:48 |
| vkmc | I added the tests | 14:48 |
| vkmc | maybe I'm the one missing something! | 14:48 |
| vkmc | correct me if I'm wrong | 14:48 |
| *** amitgandhinz has joined #openstack-zaqar | 14:50 | |
| flaper87 | vkmc: messages = {'messages': [{'ttl': 30, 'body': 'Post one!'}, ]} <- this is cheating | 14:52 |
| flaper87 | :P | 14:52 |
| flaper87 | The idea is to call post like this: | 14:53 |
| flaper87 | queue.post(dict(ttl=..,body="")) | 14:53 |
| flaper87 | or | 14:53 |
| vkmc | AH | 14:53 |
| flaper87 | queue.post(dict(ttl=..,body=""), dict(...)) | 14:53 |
| vkmc | but we are doing it like that in the API | 14:55 |
| vkmc | check L264 here https://review.openstack.org/#/c/108795/1/marconi/tests/queues/transport/wsgi/v1_1/test_messages.py | 14:55 |
| flaper87 | vkmc: yup, but that's the server test. I mean, the API requires it to be wrapped but in the library, in order to have a simpler API, we just require the message to be passed to the post call | 14:57 |
| flaper87 | at least in the higher-level API - the Queue instance, that is. | 14:57 |
| vkmc | makes sense | 14:57 |
| vkmc | ok, I'll fix it | 14:57 |
| flaper87 | vkmc: ahem... cheater... ahem | 14:58 |
| vkmc | flaper87, its kgriffs fault | 14:58 |
| flaper87 | vkmc: damn, I KNEW it was kgriffs|afk again! | 14:58 |
| vkmc | ahá | 14:59 |
| vkmc | oh and, btw, you cannot call post like you mentioned | 15:02 |
| vkmc | for several messages | 15:02 |
| vkmc | queue.post([dict(ttl=..,body=""), dict(...)]) | 15:03 |
| flaper87 | it's without the list | 15:03 |
| flaper87 | queue.post(*[dict(ttl=..,body=""), dict(...)]) | 15:03 |
| *** cpallares has joined #openstack-zaqar | 15:03 | |
| flaper87 | each message is a positional argument | 15:03 |
| vkmc | cool | 15:04 |
| *** openstackgerrit has quit IRC | 15:07 | |
| flaper87 | vkmc: can you review these doc patches? https://review.openstack.org/#/q/status:open+project:openstack/python-zaqarclient+branch:master+topic:docs,n,z | 15:09 |
| vkmc | flaper87, sure | 15:11 |
| *** kgriffs|afk is now known as kgriffs | 15:20 | |
| kgriffs | flaper87: it's not my fault. that was my evil twin. | 15:21 |
| kgriffs | flaper87, riveter: around? | 15:25 |
| vkmc | flaper87, https://review.openstack.org/#/c/129988/2 :D | 15:31 |
| vkmc | flaper87, just a nit here https://review.openstack.org/#/c/127182/5 | 16:00 |
| *** openstack has joined #openstack-zaqar | 16:02 | |
| * flaper87 back | 16:03 | |
| flaper87 | kgriffs: I'm here | 16:04 |
| flaper87 | :D | 16:04 |
| kgriffs | o/ | 16:04 |
| kgriffs | I was just wanting to through out some ideas re that patch | 16:04 |
| kgriffs | throw | 16:05 |
| flaper87 | sure thing | 16:05 |
| * kgriffs wonders why he keeps swapping homophones lately | 16:06 | |
| kgriffs | ok, for reference: https://review.openstack.org/#/c/129109/ | 16:06 |
| flaper87 | vkmc: addressed all your comments | 16:07 |
| flaper87 | vkmc: thanks :) | 16:07 |
| flaper87 | kgriffs: shoot | 16:08 |
| kgriffs | I've been thinking about this and do agree after all that the approach is fairly ugly. We are running into some limitations of the Python language here. | 16:08 |
| kgriffs | so, I have a couple alternative proposals, let me know what you think | 16:08 |
| kgriffs | option A: Revert to the previous code. | 16:09 |
| flaper87 | (as in, don't merge that patch?) | 16:09 |
| kgriffs | right | 16:09 |
| flaper87 | ok | 16:09 |
| flaper87 | next ? | 16:09 |
| * flaper87 is curious | 16:10 | |
| * flaper87 likes kgriffs ideas | 16:10 | |
| flaper87 | well, .... | 16:10 |
| flaper87 | :D | 16:10 |
| kgriffs | option B: same as above, but do remove __getattr__ and clean things up so we just implement the contract from the base class and don't forward anything else | 16:10 |
| kgriffs | (IMO the current design is confusing and inconsistent, and also does more than it should by forwarding EVERYTHING, not just the public interface contract) | 16:11 |
| kgriffs | one more | 16:11 |
| kgriffs | option C: same as above, but also use macropy to DRY up the code (let me paste a gist example) | 16:11 |
| * flaper87 doesn't know macropy | 16:12 | |
| kgriffs | https://gist.github.com/anonymous/71ba60f941446b6ae154 | 16:12 |
| kgriffs | that proxy decorator is actually a macro that is able to mess with the AST when the module is imported, before the code is compiled | 16:12 |
| * flaper87 votes for B | 16:13 | |
| flaper87 | :P | 16:13 |
| kgriffs | but yeah, that would mean having to get macropy in global-requirements and who knows if the gods will approve | 16:13 |
| flaper87 | LOOOOOOOOOL | 16:13 |
| flaper87 | yeah, the other thing is that reading: https://gist.github.com/anonymous/71ba60f941446b6ae154#file-gistfile1-py-L2-L4 | 16:14 |
| kgriffs | (yet another reason why global-requirements is a BAD IDEA) | 16:14 |
| flaper87 | is not actually intuitive and people would have to learn about macropy and stuff | 16:14 |
| kgriffs | true | 16:14 |
| kgriffs | you could mitigate that a little | 16:14 |
| kgriffs | like say @macros.proxy | 16:14 |
| kgriffs | but yeah, still a bit of a learning curve | 16:14 |
| flaper87 | yeah, in this specific case, I don't really mind the extra code there | 16:15 |
| kgriffs | ok, if you like option B then that is a precursor to C anyway if we ever decide to try it | 16:15 |
| flaper87 | awesome | 16:15 |
| flaper87 | that sounds good, we can have a proof patch for C after B lands | 16:15 |
| kgriffs | riveter: ^^^ | 16:15 |
| flaper87 | riveter: btw, thanks a lot for working on that and sorry for the confusion :) | 16:16 |
| *** sgotliv has quit IRC | 16:37 | |
| *** exploreshaifali has joined #openstack-zaqar | 16:37 | |
| *** sgotliv has joined #openstack-zaqar | 16:43 | |
| *** kgriffs is now known as kgriffs|afk | 16:44 | |
| *** openstackgerrit has joined #openstack-zaqar | 16:49 | |
| *** sgotliv has quit IRC | 16:50 | |
| *** amitgandhinz has quit IRC | 16:51 | |
| *** amitgandhinz has joined #openstack-zaqar | 17:02 | |
| *** reed has quit IRC | 17:15 | |
| *** openstackgerrit has quit IRC | 17:48 | |
| *** openstackgerrit has joined #openstack-zaqar | 17:49 | |
| *** reed has joined #openstack-zaqar | 17:54 | |
| *** jchai is now known as jchai_afk | 18:09 | |
| *** AAzza has left #openstack-zaqar | 18:15 | |
| *** exploreshaifali has quit IRC | 18:44 | |
| *** cpallares has quit IRC | 18:54 | |
| *** flwang1 has joined #openstack-zaqar | 19:00 | |
| *** jchai_afk is now known as jchai | 19:07 | |
| *** flwang1 has quit IRC | 19:25 | |
| *** AAzza_afk has joined #openstack-zaqar | 19:32 | |
| *** AAzza_afk is now known as AAzza | 19:33 | |
| *** exploreshaifali has joined #openstack-zaqar | 19:36 | |
| *** kgriffs|afk is now known as kgriffs | 19:42 | |
| * vkmc lurks | 19:48 | |
| * flaper87 lurks | 19:49 | |
| vkmc | mpanetta, ? | 19:50 |
| vkmc | kgriffs, ? | 19:50 |
| kgriffs | yo yo | 19:51 |
| vkmc | y u no lurk? | 19:51 |
| mpanetta | Hmm? | 19:51 |
| * mpanetta lurks harder | 19:51 | |
| * flaper87 lurks mpanetta lurk while kgriffs and vkmc lurk | 19:51 | |
| vkmc | lol | 19:52 |
| mpanetta | lurky lurky lurky | 19:52 |
| * flaper87 wonders where's his F*$#$^%(#$*%^#$ ligther | 19:52 | |
| mpanetta | Loose somethin flaper87? | 19:52 |
| vkmc | pipe time flaper87? | 19:53 |
| flaper87 | vkmc: thaaat's right, girl! | 19:53 |
| flaper87 | but I can't find my lighter, I must have lost it | 19:53 |
| flaper87 | damnit | 19:53 |
| flaper87 | well, I sure know how to make fire without a ligther | 19:54 |
| vkmc | use a match...? | 19:54 |
| vkmc | or maybe your good old firetorch? | 19:54 |
| flaper87 | vkmc: firetorch it is | 19:55 |
| vkmc | s/firetorch/flamethrower | 19:55 |
| flaper87 | I ran out of matches because I have (had?) a lighter | 19:55 |
| flaper87 | no no, firetorch this time | 19:55 |
| flaper87 | the flamethrower may ruin my pipe | 19:55 |
| vkmc | good, keeping things simply | 19:55 |
| vkmc | haha | 19:55 |
| vkmc | s/simply/simple | 19:56 |
| mpanetta | Eep ruined pipes no good. | 19:57 |
| *** exploreshaifali has quit IRC | 20:02 | |
| * flaper87 thinks: "Now we are talking" | 20:02 | |
| *** AAzza has left #openstack-zaqar | 20:03 | |
| * kgriffs lurks | 20:05 | |
| *** malini has quit IRC | 20:06 | |
| * flaper87 smokes | 20:09 | |
| *** flwang1 has joined #openstack-zaqar | 20:26 | |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Setup developer docs for zaqarclient https://review.openstack.org/127171 | 20:27 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Extend some docstrings with useful information https://review.openstack.org/127227 | 20:27 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Add docs for `Client` instances https://review.openstack.org/127182 | 20:27 |
| openstackgerrit | Flavio Percoco proposed a change to openstack/python-zaqarclient: Add reference docs for latest recommended client https://review.openstack.org/127215 | 20:27 |
| flaper87 | vkmc: ^ fixed the missing s | 20:28 |
| * flaper87 hates typos | 20:28 | |
| vkmc | thanks fla | 20:28 |
| flaper87 | vkmc: thank, YOU! | 20:29 |
| flaper87 | OMFG, the summit is almost 1 week way | 20:29 |
| vkmc | lets got those sweet docs merged | 20:29 |
| * flaper87 can't wait | 20:29 | |
| vkmc | oh yeeeeeeeeeeeeeeees | 20:30 |
| mpanetta | Y'all suck :P | 20:30 |
| * mpanetta wants to go to france | 20:30 | |
| flaper87 | mpanetta: I was always talking about the week of really really hard work | 20:30 |
| * flaper87 acts inocent | 20:30 | |
| vkmc | booo mpanetta, I thought you were coming too | 20:30 |
| vkmc | who is going to lurk? | 20:30 |
| vkmc | :( | 20:31 |
| mpanetta | Not this time :( | 20:31 |
| mpanetta | flaper87: Which week? :P | 20:31 |
| flaper87 | LOL | 20:34 |
| * flaper87 always regrets the week-party when it ends | 20:35 | |
| flaper87 | I mean, the summit | 20:35 |
| vkmc | I couldn't attend Icehouse and Juno | 20:39 |
| vkmc | I've been waiting for Kilo :o I'll be too sad when it ends | 20:40 |
| flwang | flaper87: do you have another meeting to discuss the summit topics? or it's done already? | 20:42 |
| vkmc | we haven't meet yet, right? we should do that | 20:43 |
| flaper87 | flwang: vkmc we haven't had it, it's all my fault | 20:44 |
| flaper87 | we can have it now if you guys can | 20:44 |
| flaper87 | I'm available now | 20:44 |
| flaper87 | vkmc: flwang kgriffs ? | 20:44 |
| flaper87 | it shouldn't take long | 20:44 |
| vkmc | I can do that | 20:44 |
| kgriffs | o/ | 20:44 |
| vkmc | malini|afk? | 20:44 |
| kgriffs | \o\ | 20:44 |
| kgriffs | /o/ | 20:44 |
| kgriffs | \o/ | 20:44 |
| flaper87 | kgriffs: are you skiing ? | 20:44 |
| flaper87 | :P | 20:44 |
| kgriffs | listening to music | 20:45 |
| flaper87 | oh ok ok :D | 20:45 |
| flaper87 | flwang: kgriffs vkmc https://etherpad.openstack.org/p/kilo-zaqar-summit-topics | 20:45 |
| flaper87 | Lets focus on the design sessions, we;ll organize the pod discussions later | 20:45 |
| kgriffs | ok. which ones are we doing design sessions for? | 20:45 |
| flaper87 | kgriffs: that we need to discuss now | 20:46 |
| flaper87 | :P | 20:46 |
| kgriffs | lol | 20:46 |
| flaper87 | The first session listed there is v2 | 20:46 |
| flaper87 | I'm a bit torn there since v2 is becoming more and more important but we still don't have any other feedback from the community | 20:46 |
| flaper87 | .. on v1 | 20:46 |
| flaper87 | Still, I think it'd be worth digging in what we think is wrong on v1.1 | 20:47 |
| flaper87 | and what major changes we see for v2 | 20:47 |
| flaper87 | regardless of whether it'll happen in Kilo or L | 20:47 |
| kgriffs | v2 is a tricky topic | 20:47 |
| flaper87 | kgriffs: right, I think I would rather talk about FIFO than v2 as a whole | 20:48 |
| kgriffs | because it pulls in the discussions around message listing vs. simple SQS job queueing style semantics, AMQP, etc. | 20:48 |
| kgriffs | FIFO | 20:48 |
| kgriffs | etc. | 20:48 |
| flaper87 | but talking about changing our FIFO story means we're considering v2 | 20:48 |
| kgriffs | just take all the controversial things, put them in a bucket, and bring them to that session | 20:48 |
| kgriffs | :p | 20:48 |
| flaper87 | kgriffs: can we pick them out of a bowl randomly? | 20:49 |
| flaper87 | :P | 20:49 |
| flaper87 | that makes it more interesting | 20:49 |
| flaper87 | ok, v2 it is | 20:49 |
| flaper87 | agreed? | 20:49 |
| vkmc | lol | 20:49 |
| flaper87 | kgriffs: we are suppose to be the drivers. Do you want to lead it? Otherwise, I'll do it | 20:50 |
| kgriffs | yeah, we should at least sit down and talk about all this stuff and find out what, if anything, in v1.x is preventing people from using the service. | 20:50 |
| vkmc | agree that v2 should be ther first discussion | 20:50 |
| flwang | what's the big move for v2? | 20:50 |
| flwang | maybe I missed something | 20:51 |
| flaper87 | flwang: there are quite few things: FIFO, get-message-by-id, listing semantics | 20:51 |
| flaper87 | etc | 20:51 |
| vkmc | we are removing the get messages by id thing | 20:51 |
| vkmc | and yeah ^ | 20:51 |
| flaper87 | ok, that's a go. We can choose who's going to lead it based on who has more time to work on it on the flight there | 20:52 |
| kgriffs | FIFO is interesting because people often don't realize that once you do listing semantics that is stateless (meaning clients track their own state), you kinda end up having to do FIFO anyway, but I digress. Lots of stuff we could talk about in that session. | 20:52 |
| kgriffs | flaper87: lol | 20:52 |
| flaper87 | oh look, I'm in Europe and kgriffs is in the US | 20:52 |
| flaper87 | TOOOO BAD! | 20:52 |
| flaper87 | :D | 20:52 |
| kgriffs | WAH?! | 20:52 |
| vkmc | haha | 20:53 |
| vkmc | kgriffs will have jetlag | 20:53 |
| flaper87 | pffffff | 20:53 |
| flaper87 | there's no such thing as jetlag | 20:53 |
| flaper87 | :P | 20:53 |
| vkmc | travelling to the future is hard for people | 20:53 |
| flaper87 | vkmc: he could travel to the past and still get here :P | 20:54 |
| flaper87 | it'll be harder | 20:54 |
| flaper87 | but well | 20:54 |
| flaper87 | ok ok | 20:54 |
| flaper87 | lets move on | 20:54 |
| flaper87 | Persistent Transports | 20:54 |
| flaper87 | vkmc: that's yours | 20:54 |
| flaper87 | I'm very interested in it | 20:54 |
| flaper87 | vkmc: also, cpallares proposed to discuss the cross-api layer too | 20:54 |
| vkmc | yeah well, currently we have support for wsgi | 20:55 |
| flaper87 | which I don't think we need for the wsgi transport but I do think we'll need for persisten transports | 20:55 |
| vkmc | it would be cool to add support for another transport too | 20:55 |
| flaper87 | vkmc: you and cpallares could work on something together there | 20:55 |
| flaper87 | or we could discuss the cross-api thing in a POD session | 20:55 |
| vkmc | that's great | 20:55 |
| kgriffs | the first thing to talk about in that discussion is requirements/use cases | 20:56 |
| flaper87 | just to avoid confussion, the cross-api thing is an implementation of a wire-like protocol | 20:56 |
| vkmc | I could ping cpallares and try to do something together | 20:56 |
| vkmc | kgriffs, +1 | 20:56 |
| kgriffs | what kind of transport we choose to do will vary widely based on our end goal - the kinds of apps we have in mind | 20:56 |
| flaper87 | kgriffs: right, although I think we've a couple of strong ones | 20:56 |
| flaper87 | Horizon+websockets, for example | 20:56 |
| kgriffs | yeah, I was just going to say we should try to align it with a project we know already wants to use Zaqar | 20:57 |
| kgriffs | one other consideration is do we do the entire API or just a subset | 20:57 |
| flaper87 | I think among the ones listed there, websocket is the one we should play with to begin with | 20:57 |
| flaper87 | but we'll figure that out there | 20:57 |
| kgriffs | or, start with a subset with the goal of getting the rest done (iterate for much win) | 20:57 |
| kgriffs | ok. let's make sure to capture all this in the agenda | 20:57 |
| vkmc | I made this list https://gist.github.com/vkmc/f69337561466c4cf4f05 | 20:58 |
| flaper87 | Lets just make sure we don't waste too much time in the introduction to those protocols so that there'll be time to discuss the implementation as well | 20:58 |
| vkmc | I haven't updated yet with the pros/cons of using a persistent approach | 20:58 |
| vkmc | (sorry flaper87, no time >.>, will do this weekend) | 20:58 |
| kgriffs | I would like to propose that all discussions (formal sessions as well as pod discussions) have an etherpad prepared beforehand | 20:58 |
| flaper87 | vkmc: no rush, I mean, the summit is right there but don't worry, really. We'll have breakfast during your session | 20:58 |
| flaper87 | :P | 20:58 |
| flaper87 | kgriffs: yup, I'm working on that | 20:59 |
| vkmc | flaper87, breakfast sounds good | 20:59 |
| flaper87 | we had them for the ATL summit | 20:59 |
| kgriffs | flaper87: how many time slots do we have? | 20:59 |
| flwang | flaper87: btw, we won't discuss the notification topic this time, we will just do it in Kilo, right? | 20:59 |
| flwang | kgriffs: 4? | 20:59 |
| flaper87 | The other thing is that I'd like us to finilize sessions with a good plan forward for that topic so we get things done | 21:00 |
| flaper87 | the idea for sessions is to discuss things we have already agreed on doing | 21:00 |
| kgriffs | flaper87: lol, yes let's just take the time to start coding. :D | 21:00 |
| flaper87 | flwang: kilo, yeah! | 21:00 |
| flaper87 | kgriffs: we have 4 slots | 21:00 |
| kgriffs | kk | 21:00 |
| flwang | kgriffs: will you be there? | 21:00 |
| kgriffs | yes | 21:01 |
| flwang | awesome | 21:01 |
| flaper87 | Integration with other services | 21:01 |
| flaper87 | I pingged Zane and asked him if he was interested in participating | 21:01 |
| flaper87 | as a Heat representative | 21:01 |
| flaper87 | he said, YES! | 21:01 |
| * flaper87 so happy | 21:01 | |
| flaper87 | :P | 21:01 |
| flaper87 | that was a formal proposal | 21:01 |
| flaper87 | We probably need someone from horizon too but I'm kinda leaning towards focusing on 1 service at a time | 21:02 |
| flaper87 | so far, Heat has shown most interest in Zaqar than other projects | 21:02 |
| kgriffs | i think having 2 would help keep us honest | 21:02 |
| kgriffs | prevent us from becoming too myopic | 21:02 |
| kgriffs | but no more than 2 | 21:02 |
| vkmc | +1 kgriffs | 21:02 |
| flaper87 | right, but we need to prepare the material for the second one | 21:03 |
| kgriffs | otherwise we don't get anything done. :p | 21:03 |
| flaper87 | we never do but lets keep that between us | 21:03 |
| flaper87 | :P | 21:03 |
| kgriffs | how about inviting horizon to the transport session? | 21:03 |
| flaper87 | mmh, I think it's too Zaqar specific | 21:03 |
| kgriffs | ok | 21:03 |
| vkmc | yeah | 21:03 |
| vkmc | we should have a intergration slot | 21:04 |
| flaper87 | kgriffs: want to lead this session? | 21:04 |
| flaper87 | I can take v2 | 21:04 |
| kgriffs | so I do think having an integration session is a good use of a time slot | 21:04 |
| vkmc | and there have horizon and heat | 21:04 |
| kgriffs | flaper87: ok | 21:04 |
| flaper87 | SOLD! | 21:04 |
| kgriffs | do you have an etherpad template you want to use? | 21:04 |
| * flaper87 writes that down before this dude changes his mind | 21:04 | |
| flwang | LOL | 21:04 |
| kgriffs | or should I just slap up some ASCII art and call it a day. ;) | 21:04 |
| flaper87 | No template yet, I'll create the pads and link them in this: https://etherpad.openstack.org/p/kilo-zaqar-summit-topics | 21:04 |
| flaper87 | I'll have that done tomorrow | 21:05 |
| flaper87 | so you all can start preparing your stuff | 21:05 |
| kgriffs | ok. I made some pads for some of the sessions I proposed but they are just rough drafts and I can port the content over. | 21:05 |
| flaper87 | We have FIFO listed there but we can discuss it in the v2 session | 21:05 |
| flaper87 | and have a more detailed POD session on FIFO if needed | 21:05 |
| flaper87 | kgriffs: kk | 21:05 |
| kgriffs | agreed | 21:05 |
| flaper87 | Docs | 21:06 |
| flaper87 | Do we need a full session or can we do a 1h hackathon ? | 21:06 |
| vkmc | hackaton sounds good | 21:06 |
| flaper87 | cool | 21:06 |
| vkmc | docs are mostly covered | 21:06 |
| vkmc | we are missing HA | 21:06 |
| flaper87 | Performance testing | 21:06 |
| flaper87 | vkmc: GTK, we should work on an etherpad to list things that are missing | 21:07 |
| vkmc | flaper87, nice | 21:07 |
| flaper87 | kgriffs: did you propose the performance session? | 21:07 |
| flaper87 | Anything specific you wanted to discuss? | 21:07 |
| kgriffs | yeah, I think until we have more usage of the service, we won't have any super users show up to a formal design session anyway to discuss docs | 21:07 |
| kgriffs | so, pod time is kewl with me on that docs one | 21:08 |
| kgriffs | flaper87: yes | 21:08 |
| kgriffs | that one I thought would be good to do with some rally peeps | 21:08 |
| flwang | kgriffs: +1 re doc topic | 21:08 |
| vkmc | sounds good | 21:08 |
| flaper87 | kgriffs: ok, can that one be merged with some other infrastructure specific topics ? | 21:08 |
| flaper87 | Missing gates ? | 21:08 |
| flaper87 | Missing tests ? | 21:08 |
| kgriffs | sure | 21:08 |
| flwang | flaper87: I think it's not a 'missing' | 21:09 |
| kgriffs | we can do one big happy infra session | 21:09 |
| flwang | it most like an enhancement :) | 21:09 |
| flaper87 | +1 for the infra session | 21:09 |
| flaper87 | flwang: yeah yeah, that's what EVERYONE says | 21:09 |
| flaper87 | :P | 21:09 |
| flwang | some small sessions can share the same slot | 21:09 |
| kgriffs | flaper87: if you could seed the pad with all the topics you'd like covered, that would be cool | 21:09 |
| flaper87 | ok, lets convert that into an infra session | 21:09 |
| flaper87 | kgriffs: will do | 21:09 |
| * flaper87 writes that task down | 21:09 | |
| flwang | kgriffs: is there any session from RAX about operations? | 21:10 |
| flaper87 | ok, we just filled our 4 slots | 21:11 |
| kgriffs | i think there was a talk submission but it wasn't accepted | 21:11 |
| kgriffs | apparently we didn't bribe the right people. ;) | 21:11 |
| kgriffs | flwang: I would be happy to discuss with you and answer questions at the summit. | 21:11 |
| flwang | kgriffs: it would be awesome | 21:12 |
| kgriffs | flwang: kk | 21:12 |
| kgriffs | sounds like a plan! | 21:12 |
| flwang | we(Catalyst IT) is going to adopt zaqar in production | 21:12 |
| kgriffs | flaper87: ok, we can do the redis and notifications at the pod | 21:12 |
| kgriffs | flwang: wow, cool! | 21:12 |
| flwang | so there are a lot of questions about the operations | 21:12 |
| flwang | given the code is there and cool :D | 21:13 |
| kgriffs | flwang: maybe we should have a pod discussion around zaqar ops? | 21:13 |
| kgriffs | discuss the needs you have, share lessons learned, etc. | 21:14 |
| flwang | it would be cool | 21:14 |
| kgriffs | flaper87: ^^^ ? | 21:14 |
| kgriffs | looks like there is a topic on the pad | 21:15 |
| kgriffs | "Rock-solid operability" | 21:15 |
| flwang | I know the user of AWS are using SQS heavily, so I'm really keen to know what's the status of Zaqar in Rax world | 21:16 |
| flwang | and recently I just opened an account of Rax cloud to experience the queue service | 21:17 |
| flwang | the GUI is so simple ;) | 21:17 |
| kgriffs | heh, yeah | 21:17 |
| flwang | kgriffs: btw, seems I can't post messages for a queue on GUI, is it? | 21:19 |
| kgriffs | flwang: no, I don't think you can. we can discuss more after this meeting. | 21:19 |
| flwang | kgriffs: okay | 21:19 |
| openstackgerrit | A change was merged to openstack/python-zaqarclient: Setup developer docs for zaqarclient https://review.openstack.org/127171 | 21:19 |
| flwang | where is our PTL? | 21:20 |
| * flaper87 is back | 21:22 | |
| flaper87 | POD discussion for ops sounds good to me | 21:22 |
| openstackgerrit | A change was merged to openstack/python-zaqarclient: Add docs for `Client` instances https://review.openstack.org/127182 | 21:23 |
| kgriffs | flaper87: how are we scheduling the Pod sessions? | 21:23 |
| flaper87 | kgriffs: we aren't | 21:24 |
| flaper87 | :P | 21:24 |
| kgriffs | heh | 21:24 |
| flaper87 | jokes apart, I'll start working on that tomorrow | 21:24 |
| flaper87 | because we need to make sure folks have time to go to other sessions | 21:24 |
| kgriffs | I was just thinking last summit it was kinda hard to get everyone together via text messages and such. | 21:24 |
| flaper87 | Assuming there's internet access, we can use G+ | 21:24 |
| flaper87 | to send messages | 21:25 |
| vkmc | check the app | 21:25 |
| vkmc | there is a messaging system there | 21:25 |
| vkmc | I haven't tried it yet though | 21:25 |
| kgriffs | oh boy | 21:25 |
| flaper87 | that app never works | 21:25 |
| flaper87 | T_T | 21:25 |
| flwang | flaper87: I suggest use wechat :) | 21:25 |
| * kgriffs gets ready to post a XSS attack | 21:26 | |
| flwang | then we can create a group to talk in real time :) | 21:26 |
| kgriffs | that would be cool | 21:26 |
| flwang | please search wechat in your app store | 21:26 |
| * vkmc looks for wechat | 21:26 | |
| flwang | and create an account, then we can create a lovely group, why not have a try? | 21:26 |
| vkmc | wow, it needs access for everything | 21:27 |
| flaper87 | flwang: is there really a wechat app ? | 21:27 |
| flwang | I'm seriously :D | 21:27 |
| * kgriffs downloading | 21:28 | |
| kgriffs | this would be great | 21:28 |
| flaper87 | what about simple telegram/whatsapp ? | 21:28 |
| kgriffs | we can use it to organize discussions, dinner groups, etc. | 21:28 |
| kgriffs | hazing of flaper87 | 21:28 |
| kgriffs | you know, the usual stuff | 21:28 |
| flaper87 | LOOOOOOOOOOOOOOOL | 21:28 |
| vkmc | whatsapp sounds good too | 21:28 |
| flaper87 | I'm sure you all have whatsapp | 21:29 |
| flwang | flaper87: I'm OK :D | 21:29 |
| flwang | but the truth is I don't have whatsapp :) | 21:29 |
| flwang | does it support group? | 21:29 |
| flaper87 | flwang: LIAAAAAAAAAAAAAR | 21:29 |
| flaper87 | flwang: it does | 21:29 |
| flwang | cool | 21:30 |
| flaper87 | flwang: don't act as if you don't know | 21:30 |
| flaper87 | I'm sure you wast half of your day there | 21:30 |
| flwang | the bad thing is I forgot to take my phone today, damn it | 21:31 |
| kgriffs | does whatsapp let you use "fun animated stickers"? | 21:31 |
| kgriffs | cause that is a deal breaker for me. ;) | 21:32 |
| kgriffs | ooh, cool. "Friend Radar". wechat actually looks pretty slick. anyway, whatever you all want to use, I'm cool with. | 21:32 |
| flwang | kgriffs: yes | 21:33 |
| flwang | it has the Radar | 21:33 |
| flwang | in the same area, if there are some other guys open the Radar, you can find him/her | 21:33 |
| flwang | it's super cool when you in a party instead of adding them one by one :) | 21:34 |
| flwang | i mean with their id | 21:34 |
| vkmc | kgriffs, do you have whatsapp? | 21:34 |
| vkmc | flaper87 already created a group haha | 21:34 |
| *** sriram has quit IRC | 21:34 | |
| flaper87 | kgriffs: flwang your mobile numbers, private message, now! | 21:35 |
| flwang | i'm wechat id is OpenStacker, please add me if you're using it :D | 21:35 |
| flwang | s/i'm/my/ | 21:35 |
| flwang | my NZ mobile number is +64 021 0832 6348 | 21:35 |
| flwang | kgriffs: please tell me your wechat id if you have installed and have any fun | 21:36 |
| *** amitgandhinz has quit IRC | 21:39 | |
| kgriffs | I just set my wechat id to "kgriffs" | 21:40 |
| flwang | kgriffs: awesome, I will catch you later | 21:41 |
| * flaper87 confused | 21:41 | |
| * flaper87 thought we were using whatsapp | 21:41 | |
| flwang | flaper87: haha | 21:41 |
| flwang | we could | 21:41 |
| flwang | but that doesn't impact kgriffs and I have some secret :D | 21:41 |
| kgriffs | lol | 21:43 |
| kgriffs | what are we using again? | 21:43 |
| * kgriffs downloads ALL THE THINGS! | 21:43 | |
| vkmc | ICQ | 21:43 |
| kgriffs | AOL | 21:43 |
| flaper87 | ok, flwang you are the only one missing in our whatsapp group | 21:44 |
| * flaper87 is using ZAQARRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRR | 21:44 | |
| flaper87 | you should all be ashamed | 21:44 |
| vkmc | ok.. lets use zaqar | 21:45 |
| vkmc | is too sexy to use | 21:46 |
| * flwang is really shamed without a phone now :) | 21:46 | |
| flwang | I really hope to have a whatsapp ubuntu version :) | 21:47 |
| flaper87 | kk, I'm off guys! ttyt | 21:48 |
| vkmc | flaper87, ttfn, enjoy the rest of your evening! | 21:48 |
| kgriffs | let's just sell Zaqar to Facebook and retire young. | 21:56 |
| kgriffs | oh wait, open source. | 21:57 |
| kgriffs | oooooops | 21:57 |
| *** mpanetta has quit IRC | 22:02 | |
| vkmc | kgriffs, if you have a moment later, could you help me brainstorm some other use cases for persistent transport? | 22:03 |
| vkmc | or somebody else :) | 22:05 |
| *** flwang has quit IRC | 22:05 | |
| kgriffs | vkmc: sure, is now good? | 22:06 |
| kgriffs | I can stick around for 5-10 more minutes | 22:06 |
| kgriffs | otherwise, you can ping me tomorrow | 22:06 |
| vkmc | its good now yeah! | 22:06 |
| *** flwang has joined #openstack-zaqar | 22:06 | |
| vkmc | as flaper87 mentioned, one of the cases is for Horizon | 22:07 |
| vkmc | with websockets | 22:07 |
| vkmc | I checked the use case list for other projects but I'm not sure if a persistent connection would make any difference | 22:08 |
| *** sgotliv has joined #openstack-zaqar | 22:08 | |
| kgriffs | honestly, I don't think there would be much different between websockets and long polling | 22:08 |
| vkmc | https://etherpad.openstack.org/p/zaqar-integrated-projects-use-cases | 22:08 |
| kgriffs | web sockets are good for things that are *actually* realtime such as streaming and shared whiteboards | 22:08 |
| kgriffs | there is one exception there | 22:09 |
| kgriffs | if you are pushing a TON of events and the backend data store can keep up, then a persistent connection may help | 22:09 |
| kgriffs | but really, with websockets you are probably talking a private cloud deployment since scaling persistent connections is a PITA (and gets expensive fast) | 22:10 |
| kgriffs | that social media company | 22:11 |
| kgriffs | I think it is mentioned here | 22:11 |
| kgriffs | https://etherpad.openstack.org/p/zaqar-overcloud-use-cases | 22:11 |
| kgriffs | they wanted websockets (specifically, socket.io) | 22:11 |
| vkmc | aha.. makes sense | 22:11 |
| kgriffs | but after we chatted they admitted that realistically, you can get what you need most of the time with keep-alive; other things are more likely to be a bottleneck. | 22:12 |
| kgriffs | that being said, there is something about the "wow!" factor if you can do web sockets | 22:12 |
| kgriffs | It makes a great demo | 22:13 |
| vkmc | yeah that's true | 22:13 |
| vkmc | so probably its not about what we can do with it | 22:13 |
| vkmc | but if we can stick with the trend | 22:13 |
| vkmc | everyone is leaning towards websockets nowadays | 22:13 |
| kgriffs | I'm not saying we shouldn't do it, but just that we need to understand what the goals are. | 22:14 |
| vkmc | yeah, that is what I want to understand too | 22:14 |
| vkmc | because 'persistent connection' sounds great | 22:14 |
| vkmc | but, why do we want that? | 22:14 |
| vkmc | how is Zaqar going to be improved with that? | 22:14 |
| kgriffs | right | 22:14 |
| *** jchai has quit IRC | 22:14 | |
| kgriffs | HTTP 2.0 will put a lot of this debate to rest | 22:15 |
| kgriffs | but in the meantime, it may be worth doing websocket. | 22:15 |
| vkmc | ok.. | 22:16 |
| kgriffs | it will make latency a little better, and also everyone can get excited about it | 22:16 |
| kgriffs | sorry, don't mean to sound cynical | 22:16 |
| kgriffs | i mean, it is a great selling point | 22:16 |
| kgriffs | no doubt | 22:16 |
| vkmc | haha not at all! its useful criticism | 22:16 |
| vkmc | so we know what to expect | 22:17 |
| kgriffs | now, if you are hosting using pypy and the redis driver, you might notice a bigger difference in throughput or whatever by using a persistent transport | 22:17 |
| kgriffs | in any case, I think the two use cases are: | 22:18 |
| kgriffs | control surfaces (have to communicate over the internetz but are concerned about latency for whatever reason) | 22:18 |
| kgriffs | and | 22:18 |
| kgriffs | high-throughput use cases where they need a multi-tenant, authenticated solution so something like Kafka doesn't work | 22:19 |
| kgriffs | anyway, that's my $0.02 | 22:19 |
| kgriffs | I'm sure teh Flavio well have more things to add | 22:20 |
| kgriffs | final thought | 22:20 |
| vkmc | I'll ping him tomorrow | 22:20 |
| vkmc | I also mentioned other alternatives, like WAMP and AMQP | 22:20 |
| kgriffs | if we go this way, we should think about whether straight websockets is ok or if we need something that can fallback to other things ala socket.io | 22:20 |
| vkmc | exactly yes | 22:20 |
| kgriffs | ah, yeah. WAMP is definitely interesting. | 22:21 |
| vkmc | it has a lot ot things we won't use IMO | 22:21 |
| kgriffs | yeah | 22:22 |
| kgriffs | http://wamp.ws/compared/ | 22:22 |
| vkmc | yeah | 22:22 |
| kgriffs | STOMP is another one we should consider. | 22:22 |
| kgriffs | you can pretty much do anything you want over websocket | 22:22 |
| kgriffs | I think we should optimize for | 22:22 |
| vkmc | isn't STOMP only text? that might be a bit constraining | 22:23 |
| kgriffs | so is HTTP | 22:23 |
| kgriffs | ;) | 22:23 |
| vkmc | true dat | 22:24 |
| kgriffs | I just bring it up because it is generic enough and alike enough to HTTP that it may be the simplest thing to map to our API | 22:24 |
| vkmc | yeah | 22:24 |
| kgriffs | actually, I guess HTTP isn't strictly "text"... | 22:24 |
| kgriffs | hmmm | 22:24 |
| * kgriffs looks at the STOMP spec | 22:24 | |
| kgriffs | "Otherwise, the receiver SHOULD consider the body to be a binary blob." | 22:26 |
| vkmc | ah well | 22:26 |
| kgriffs | anyway | 22:26 |
| vkmc | I also looked into crossbar.io http://crossbar.io/howitworks/ | 22:26 |
| kgriffs | getting off into the weeds | 22:26 |
| vkmc | is an implementation of WAMP | 22:26 |
| kgriffs | oic | 22:26 |
| vkmc | but I heard really good comments | 22:26 |
| kgriffs | hmm | 22:27 |
| kgriffs | you know, to be honest, I wonder if going down this path is just going to confuse the mission of zaqar again | 22:28 |
| kgriffs | if you want WAMP, why wouldn't you just use crossbar.io? | 22:28 |
| vkmc | yeah | 22:28 |
| vkmc | we have to be careful with this | 22:29 |
| kgriffs | we could say "we offer a multi-tenant and auth layer on top of all those things" | 22:29 |
| kgriffs | but I don't think that is every going to work. the backends are too different | 22:29 |
| vkmc | I'm mostly convinced that we should go with websockets or raw tcp | 22:29 |
| vkmc | because those are more generic and one might expect that Zaqar clients use those transports | 22:30 |
| vkmc | idk honestly, its a tough call | 22:31 |
| kgriffs | yeah, can you make a note about that session that we should be really clear on zaqar's mission and make sure the transport choice aligns with that? | 22:33 |
| vkmc | yeah | 22:34 |
| vkmc | of course | 22:34 |
| vkmc | that should be our north | 22:34 |
| kgriffs | I'm thinking we will have evaluation criteria like | 22:34 |
| kgriffs | 1. does it fit in our mission of multi-tenant, HA, scalable, durable messaging? | 22:34 |
| kgriffs | 2. how hard will it be to map the semantics to our API? | 22:35 |
| kgriffs | 3. How much client library support is out there? | 22:35 |
| kgriffs | 4. Will it make a super awesome demo at meetups (srsly. buzz FTW!) | 22:36 |
| kgriffs | anyway, you get the idea. | 22:36 |
| kgriffs | </soapbox> | 22:36 |
| vkmc | 5. which should be the example use cases? | 22:36 |
| vkmc | I do | 22:36 |
| kgriffs | kewl | 22:36 |
| vkmc | thanks kgriffs, its always useful your insight | 22:37 |
| kgriffs | heh, well I'm sometimes a little too opinionated I'm afraid. :p | 22:37 |
| vkmc | nah, you are fine :) | 22:38 |
| kgriffs | :) | 22:42 |
| *** amitgandhinz has joined #openstack-zaqar | 22:49 | |
| riveter | kgriffs, got a minute? | 22:53 |
| kgriffs | sure | 22:54 |
| *** amitgandhinz has quit IRC | 22:54 | |
| riveter | so the idea now is to remove __getattr__ and just put any methods that are identical between the 3 metaclasses into RoutingController | 22:54 |
| riveter | ? | 22:54 |
| riveter | *metacontrollers, sorry | 22:56 |
| kgriffs | for the most part. let me clarify my proposal and we can see if it makes sense. :p | 22:58 |
| riveter | shoot | 22:58 |
| kgriffs | so, given that I think we are all starting to think it is pretty hacky to try and DRY this up without using something sneaky like macropy | 22:59 |
| riveter | not familiar with macropy | 23:00 |
| kgriffs | let's just embrace what Python does instead of trying to be so clever. By which I mean, let's not even try to do this __getattr__ thing because it actually doesn't give us a lot of value IMO. | 23:01 |
| openstackgerrit | A change was merged to openstack/python-zaqarclient: Add reference docs for latest recommended client https://review.openstack.org/127215 | 23:01 |
| riveter | yeah | 23:01 |
| kgriffs | (since the calling conventions and responses when lookups fail vary so widely) | 23:01 |
| kgriffs | also, it actually does more than we want | 23:01 |
| kgriffs | we only want to implement the base class interface, but we end up actually forwarding everything under the sun | 23:01 |
| kgriffs | so, here is what I was thinking to do instead | 23:02 |
| kgriffs | if you look in zaqar/queues/storage/base.py | 23:02 |
| kgriffs | the interfaces for each controller that we want to forward are defined | 23:03 |
| riveter | yeah | 23:03 |
| riveter | I looked at this but decided not to mess with it | 23:03 |
| kgriffs | so, just inherit from those base classes like the real drivers do | 23:04 |
| riveter | mm, looks like most of the base class methods right now are not implemented | 23:05 |
| riveter | you mean to move the implementations there? | 23:05 |
| kgriffs | but the bodies will just be "do a lookup, call the target's same method. If no lookup, return a default answer or raise an error" | 23:05 |
| openstackgerrit | A change was merged to openstack/python-zaqarclient: Extend some docstrings with useful information https://review.openstack.org/127227 | 23:05 |
| kgriffs | hmmm | 23:06 |
| riveter | sorry, not sure I follow | 23:06 |
| kgriffs | let me look | 23:06 |
| riveter | that's what the metacontrollers are already doing | 23:06 |
| kgriffs | i guess that is my point | 23:07 |
| kgriffs | what is the point of keeping __getattr__ around? | 23:07 |
| riveter | yeah, I'm not arguing with that | 23:07 |
| riveter | I'm just not sure what you're suggesting to put in the base classes | 23:07 |
| kgriffs | oh, in base.py you would not change anything | 23:07 |
| kgriffs | in pooling.py | 23:08 |
| riveter | so how does inheriting from them help? | 23:08 |
| kgriffs | you would change MessageController to inherit from zaqar.queues.storage.Message (as do the redis, mongo, and sqla drivers) | 23:09 |
| kgriffs | riveter: let me try to answer that | 23:10 |
| kgriffs | although Python's inheritance doesn't enforce implementation of the interface as strictly as say C++ or Java, it does make sure you don't forget to add a method in the child classes when the parent changes | 23:11 |
| kgriffs | it also serves as a reminder when reading the code what the contract is | 23:11 |
| riveter | mm, ok | 23:11 |
| riveter | so we're not looking at removing the duplicated code at all | 23:11 |
| kgriffs | I suspect that some of the method signatures have gotten out of sync with the base class interface as well, in subtle ways. | 23:11 |
| riveter | yeah, I noticed a few cases of that | 23:12 |
| kgriffs | inheriting won't catch that, but as part of this patch we should make sure the args and kwargs are in sync | 23:12 |
| kgriffs | so, basically | 23:12 |
| riveter | spent some time looking for a subtle reason, then decided maybe there wasn't one :) | 23:12 |
| kgriffs | 1. remove __getattr__ | 23:12 |
| riveter | 2. make pooling controllers inherit from base classes | 23:13 |
| kgriffs | right | 23:13 |
| riveter | 3. check that method signatures are consistent | 23:13 |
| kgriffs | you can also inherit from routing controller | 23:13 |
| riveter | can I suggest a 4? | 23:13 |
| riveter | yeah | 23:14 |
| kgriffs | but now all it does is factor out the __init__ line | 23:14 |
| kgriffs | self._lookup = self._pool_catalog.lookup | 23:14 |
| riveter | yeah, the __init__ methods could actually be removed entirely, they're identical & could be done in RoutingC | 23:14 |
| kgriffs | right. I don't see anything right off that could be factored out. | 23:14 |
| kgriffs | makes RoutingController pretty simple | 23:15 |
| riveter | yeah | 23:15 |
| riveter | one thought though? | 23:15 |
| riveter | the really repetitive code all looks like | 23:16 |
| riveter | target = self._lookup(queue, project) | 23:16 |
| riveter | if target: control = target.message_controller | 23:17 |
| riveter | could catalog just provide a lookup_message_controller method? | 23:17 |
| riveter | and similarly for queue and claim? | 23:17 |
| kgriffs | sure | 23:18 |
| riveter | it'd trim one line out of every method without being weird or hitting performance | 23:18 |
| kgriffs | sounds good! | 23:18 |
| riveter | ok | 23:18 |
| riveter | um | 23:18 |
| kgriffs | so I know this is sort of the opposite of what this bug was going to do | 23:19 |
| riveter | is this just patch-set 3 for the same review process? | 23:19 |
| kgriffs | but I think I would rather choose "clean and simple" rather than "DRY and super hacky" | 23:19 |
| riveter | or do I start a new branch, or something? | 23:19 |
| kgriffs | riveter: good question | 23:19 |
| riveter | yeah, that's probably just as well | 23:19 |
| riveter | clever code is more fun to write than to maintain :) | 23:19 |
| kgriffs | it is probably easiest to start a new branch and abandon the old one, since this approach is quite different | 23:19 |
| riveter | ok | 23:20 |
| riveter | still referring to the same bug though? | 23:20 |
| kgriffs | riveter: so, after all this is done, if we want to still try to DRY things a little more we could use macropy | 23:20 |
| kgriffs | but idk if it is worth it | 23:20 |
| kgriffs | https://gist.github.com/anonymous/71ba60f941446b6ae154 | 23:20 |
| riveter | honestly, if the catalog methods go in there won't be that much repetition | 23:20 |
| kgriffs | that @proxy is actually a macro that changes the AST when the module is imported | 23:20 |
| riveter | ooh | 23:20 |
| kgriffs | riveter: yeah, good point | 23:20 |
| riveter | shiny | 23:20 |
| kgriffs | anyway, I gotta run | 23:21 |
| riveter | ok, thanks | 23:21 |
| kgriffs | sure. | 23:21 |
| riveter | I'm not around much for the next week, but will do this first week of nov | 23:21 |
| kgriffs | OK | 23:21 |
| kgriffs | sorry for the change of direction on this | 23:22 |
| kgriffs | sometimes you have to give something a try before you know whether it is a good idea or not. :p | 23:22 |
| riveter | no worries :) | 23:22 |
| riveter | cheers | 23:22 |
| kgriffs | thanks! | 23:22 |
| kgriffs | take care | 23:22 |
| kgriffs | o/ | 23:22 |
| riveter | \o | 23:23 |
| *** X019 has joined #openstack-zaqar | 23:31 | |
| *** sgotliv has quit IRC | 23:39 | |
| *** gmann has left #openstack-zaqar | 23:39 | |
| *** kgriffs is now known as kgriffs|afk | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!