Wednesday, 2015-05-13

*** salv-orlando has joined #openstack-api00:15
*** annegentle has joined #openstack-api00:17
*** annegentle has quit IRC00:23
*** salv-orlando has quit IRC00:26
*** annegentle has joined #openstack-api00:33
*** fzdarsky_ has joined #openstack-api01:42
*** fzdarsky has quit IRC01:46
*** annegentle has quit IRC02:07
*** annegentle has joined #openstack-api02:40
*** annegentle has quit IRC02:41
*** annegentle has joined #openstack-api03:10
*** salv-orlando has joined #openstack-api03:17
*** annegentle has quit IRC03:20
openstackgerritAlex Xu proposed openstack/api-wg: Clarify the return code when server have hard-code length limit
openstackgerritAlex Xu proposed openstack/api-wg: Clarify the return code when server have hard-code length limit
*** salv-orlando has quit IRC03:42
*** sigmavirus24 is now known as sigmavirus24_awa04:06
*** salv-orlando has joined #openstack-api04:10
*** fzdarsky_ has quit IRC04:22
*** fzdarsky_ has joined #openstack-api04:22
*** fzdarsky_ is now known as fzdarsky04:22
*** salv-orlando has quit IRC04:25
*** salv-orlando has joined #openstack-api04:32
*** salv-orlando has quit IRC04:33
*** fzdarsky has quit IRC05:33
*** fzdarsky has joined #openstack-api06:34
*** smallbig has left #openstack-api06:49
*** fifieldt has joined #openstack-api06:56
*** fifieldt has quit IRC06:56
*** woodster_ has quit IRC07:00
*** e0ne has joined #openstack-api08:25
*** e0ne has quit IRC08:28
*** e0ne has joined #openstack-api08:30
*** e0ne has quit IRC08:31
*** fzdarsky_ has joined #openstack-api08:50
*** e0ne has joined #openstack-api09:22
*** e0ne is now known as e0ne_09:36
*** e0ne_ is now known as e0ne09:38
*** e0ne is now known as e0ne_10:25
*** e0ne_ is now known as e0ne10:26
*** e0ne is now known as e0ne_10:39
*** e0ne_ is now known as e0ne10:40
*** cdent has joined #openstack-api11:13
*** e0ne is now known as e0ne_11:26
*** salv-orlando has joined #openstack-api11:28
*** salv-orlando has quit IRC11:37
*** salv-orlando has joined #openstack-api11:43
*** woodster_ has joined #openstack-api11:59
*** e0ne_ is now known as e0ne12:08
*** annegentle has joined #openstack-api12:34
*** salv-orlando has quit IRC12:39
*** fzdarsky_ has quit IRC12:44
*** annegentle has quit IRC12:45
*** annegentle has joined #openstack-api12:46
*** annegentle has quit IRC13:06
*** annegentle has joined #openstack-api13:07
*** sigmavirus24_awa is now known as sigmavirus2413:37
sdaguecdent: you around?14:03
sdagueI wanted to spitball about what should be added to the http section to fill it out more completely14:03
cdentyeah, did you see the updates I made this morning?14:05
cdentsorry forgot my sdague ^14:05
sdagueI saw a POST / PUT one14:06
cdentI made some additional comments on your "background" review14:06
sdagueok cool, I didn't look yet14:07
sdagueyeh, my concern about sending people directly to the rfc ... is that's how we got the massive missuse of 50114:08
sdaguethe rfc is like legal code14:08
sdagueit doesn't exist in a vacuum, precedents are also important14:08
sdagueso better to cycle back around if there is vaguery14:08
cdentwell sure, but if during that cycling around you take some time to review the rfc[s] then all the better for the ensuing conversations14:10
cdentI don't think we want these guidelines to allow people to not have at least glanced at the rfcs14:10
cdentand that for things like 5xx we need catch all rules of thumb (don't use 5xx unless you are really sure you know what you're doing)14:13
cdenta server should produce 500, not an app14:13
cdentand unfortunately we tend to think of the api services as servers14:14
sdagueso, I was thinking about this a bit differently, which is that instead of this being a patch on top of the rfc to clarify it for our use case, it is really the entry point document and should uplift the cliff notes of http use in OpenStack14:14
sdaguebecause, honestly, most people looking at this are going to be developers that aren't all that familiar just trying to add another feature to a project14:15
cdentyeah, and we should slow down the process of adding features as much as possible in the name of quality ;)14:16
cdent(not actually serious, but I could almost be convinced to be so)14:16
sdagueyeh, well, I don't think that's the job of this spec :)14:16
cdentI reckon the job of the spec is to make sure the apis are good14:17
cdentand the challenge is that there are differing levels of good within and without the group14:17
cdentand different thoughts on the strategies14:17
cdentI think the guidelines are aspirational14:17
cdentothers think they should validate the current state and use that as the basis for consistency14:18
cdentgiven the amount of "oh my god it does what!" in some of the apis, I'd much prefer that the guidelines drive change14:18
cdentunfortunate we've already got so much other change that needs to happen, and that endless drive for features that it is likely that too much aspiration will just get washed away14:19
cdentor ignored14:19
cdentI've tried several times to get the "what are these guidelines for" or "what angle are we coming at this from" question resolved several times with little success14:20
cdentI think, as with so many things, it's incremental change in the belly of gerrit that rules the day. Thus my encouragement for you to put your historical context front and center at the top of the document14:21
elmikoi thought we've discussed the aspirational notion of the guidelines, that they aren't supposed to be hard rules14:22
sdagueso... that's my rought brain storm of an outline14:23
elmikoi think the process of improving the apis across openstack is, by nature, a slow process that will involve getting more folks informed about the guidelines14:23
elmikoi've seen small improvements with more folks wanting to investigate the guidelines and take action based on the advice14:24
sdaguesure, but there is even a lot of best practice that's currently in the heads of folks, and mostly writing that down, has a big impact14:24
cdentelmiko: I guess there is agreement that they are aspirational but not sure if there is agreement on the height of the aspiration14:24
sdaguethere are a lot of lurkers that just read these things and run with them14:24
elmikocdent: good point14:24
cdentsdague: that seems a good outline14:25
elmikosdague: looking at the etherpad, do you think we should keep this as one guideline, because i can see 2 (one for response codes, and one for method usage)14:25
cdentI think your last point, though, is at the core of why so much of these discussion are troubling for me14:25
sdaguecdent: which point?14:26
cdentif we model the requests as raw then of course they will work with most libraries14:26
cdentunless those libraries can't do http14:26
cdentand if those libraries can't do http we should be using them14:26
cdent(e.g. wsme)14:26
sdagueso... by libraries, I don't mean ours, I mean things like client libs14:27
sdaguelike the DELETE with body question14:27
sdaguewhere it's an odd edge case14:27
sdagueand you shouldn't assume that DELETE calls with a body are going to work with all libraries and proxies14:27
cdentI didn't understand why that was edge case at all. It was just wrong.14:27
sdaguecdent: yeh, though not directly verbotten in the spec14:28
sdagueI did reread a bunch, and whoever called that out, was right14:28
cdentindeed, why your point about "writing down the wisdom contained in heads" makes good sense14:28
sdaguebut it was a context issue14:28
sdagueI agree that if we have libraries that massively violate http in our universe we should call them out14:29
sdagueanyway, I need to run and do errands then lunch, I mostly want to figure out if that outline looked sane, and if so, get people to (ircnick) bullets they want to write, and we can patch flood in a bunch of stuff and flesh this all out quickly14:30
sdaguebecause as a divide and conquer we could get a bunch of stuff in, and that would be great14:30
cdentit is sane14:30
cdentthanks for putting it together14:31
cdent(shame about that mapping to CRUD thing though :P )14:31
cdentI hope its understood that my stance as an HTTP idealist is for the sake of balance, not because I'm (only) a total jerk.14:33
sdaguecdent: heh14:42
cdentthere's plenty of "let's just get on with it" so we need one or two "ingrates, purify thineselves"14:43
sdaguedude, I'll say again, I'm super happy you are in the middle of these conversations cdent14:48
cdentDitto, I'm glad you are hopping in to the fray with enthusiasm. I really don't know how you manage the time slicing you seem to do.14:49
elmikocdent: so, are you talking about more emphasis on evangelism as opposed to just creating guidelines?14:50
*** e0ne is now known as e0ne_14:50
* cdent thinks14:51
*** e0ne_ is now known as e0ne14:52
cdentI guess in fantasy world all of OpenStack would include considerably more "active leadership" which is probably another word for "evangelism".14:52
cdentThere's an extent in which this entire project is a huge boondoggle for levelling people up in the understanding of all kinds of stuff.14:52
elmikoi can see that14:52
cdentAnd if that's the case then it would be great to take advantage of it and churn out folk who are teh ossum14:53
elmikoimo, i think we could do more with the liaisons that i think could help drive the change14:53
cdentOf course all that assumes that those of us who are actively invested in that idea are not spread very thinly across a giant map.14:54
elmikogood point14:55
cdentand/or that our funders will let us devote the time14:55
elmikoand that's part of the issue, everyone has differing levels of time they can commit to seemingly "extraneous" activities14:56
*** e0ne is now known as e0ne_15:02
cdentI feel like we're gathering some good momentum. In large part because people are forming relationships that allow them to actually understand one another a bit. From there all kinds of awesome can happen so it will be interesting to see what happens over the next few months.15:04
elmikocdent, sdague, should we bring this etherpad up at the meeting tonight to get more folks onboard?15:04
cdentAfter that bit of optimism I need some coffee. biab15:04
cdentelmiko: yes15:04
elmikocdent: agreed about relationships =)15:05
cdenttonight is the midnight meeting, yeah? can't make it :(15:05
* cdent coffees15:05
elmikocdent: no worries, i'll bring it up. i just want to make sure i understand the intent.15:06
*** salv-orlando has joined #openstack-api15:22
*** salv-orlando has quit IRC15:41
*** e0ne_ is now known as e0ne15:43
*** fzdarsky has quit IRC16:44
*** cdent has quit IRC17:30
*** salv-orlando has joined #openstack-api17:30
*** salv-orlando has quit IRC17:44
sdagueelmiko: when are meetings? I will probably not be able to make today because of childcare things17:55
*** e0ne is now known as e0ne_18:00
*** annegentle has quit IRC18:04
*** e0ne_ is now known as e0ne18:04
*** e0ne has quit IRC18:06
elmikosdague: meeting is at 00:00UTC (8pm eastern)18:12
elmikosdague: i just wanted to make i understood you, so essentially i'll show the etherpad and see if anyone wants to volunteer to write a section?18:13
sdagueyeh, that will almost never work for me due to family things18:14
sdaguebut that's fine18:14
sdagueyes, my intent is to "fill out" the http section to get everything out of people's heads and onto paper, hopefully quickly18:14
sdaguethe etherpad is an attempt to outline what I think we need, and to make it managable basically have people submit content for like 1 bullet per patch, and split it up on any authors that want to contribute18:15
elmikook, cool. sounds good18:16
elmikoi'll shop it around and see if we can get some interest18:16
ryansbelmiko: I'm interested but can't make the meeting tonight either :(18:23
elmikoryansb: if you are interested in handling one of the topics in the etherpad just mark your irc nick next to it18:23
elmikoand no worries, i'm sure we'll get to talk about this more next week =)18:24
ryansbelmiko: indeed, we have actual *physical* work sessions. Weird18:24
ryansbam I silly and missed the etherpad link somewhere?18:26
ryansbah, never mind
sdagueelmiko: I can write more word things in the etherpad to set the tone18:34
elmikosdague: if you have a spare moment that would be awesome, but if not i think i get where you are going.18:36
sdagueelmiko: ok, more words in the etherpad18:43
*** e0ne has joined #openstack-api18:45
elmikosdague: +1 thanks18:48
*** annegentle has joined #openstack-api19:00
*** e0ne has quit IRC19:32
*** e0ne has joined #openstack-api19:33
* sigmavirus24 wont' make it to the meeting tonight, if we're still having one19:37
elmikoi'll be there, but we'll see if we have a quorum19:40
elmikosigmavirus24: thanks for the heads up19:40
*** e0ne has quit IRC19:43
*** e0ne has joined #openstack-api19:59
*** annegentle has quit IRC20:39
*** annegentle has joined #openstack-api20:39
*** annegentle has quit IRC21:32
*** stevelle has quit IRC21:41
etoewsi just caught up on this conversation22:08
etoewsadded a link to sdague's etherpad for the liberty etherpad for the api wg sessions.22:08
etoewsi still can't make it to the meeting tonight though :(22:10
elmikoit sounds like it might be fairly light though22:10
etoews*adds meeting times to list of things to discuss at the summit*22:11
*** salv-orlando has joined #openstack-api22:15
*** salv-orlando has quit IRC22:15
*** e0ne has quit IRC22:28
*** openstackgerrit has quit IRC22:47
*** openstackgerrit has joined #openstack-api22:49

Generated by 2.14.0 by Marius Gedminas - find it at!