Tuesday, 2015-07-28

*** vishwanathj has quit IRC00:05
*** salv-orlando has joined #openstack-api00:33
*** salv-orlando has quit IRC00:37
*** openstackgerrit has quit IRC00:46
*** openstackgerrit has joined #openstack-api00:47
*** Apoorva has quit IRC01:14
*** terrylhowe_ has joined #openstack-api01:32
*** terrylhowe_ has left #openstack-api01:33
*** woodster_ has quit IRC01:34
*** salv-orlando has joined #openstack-api01:44
*** salv-orlando has quit IRC01:48
*** dolphm has quit IRC02:07
*** sigmavirus24 has quit IRC02:07
*** dolphm has joined #openstack-api02:09
*** dolphm has quit IRC02:09
*** dolphm has joined #openstack-api02:10
*** Apoorva has joined #openstack-api03:07
*** Apoorva has quit IRC04:32
*** e0ne has joined #openstack-api05:14
*** e0ne has quit IRC05:42
*** e0ne has joined #openstack-api05:44
*** e0ne has quit IRC05:51
*** salv-orlando has joined #openstack-api05:55
*** e0ne has joined #openstack-api05:55
*** e0ne has quit IRC06:13
*** vishwanathj has joined #openstack-api06:28
*** salv-orlando has quit IRC06:59
*** cdent has joined #openstack-api07:02
*** e0ne has joined #openstack-api07:37
*** alex_klimov has joined #openstack-api07:38
*** salv-orlando has joined #openstack-api07:50
*** lucasagomes has joined #openstack-api08:08
*** vishwanathj has quit IRC08:27
*** vishwanathj has joined #openstack-api08:28
*** vishwanathj has quit IRC08:39
*** vishwanathj has joined #openstack-api08:53
*** vishwanathj has quit IRC09:10
*** vishwanathj has joined #openstack-api09:12
*** alex_klimov has quit IRC10:29
*** salv-orl_ has joined #openstack-api10:34
*** salv-orlando has quit IRC10:37
*** salv-orl_ has quit IRC10:47
*** salv-orlando has joined #openstack-api11:00
*** salv-orl_ has joined #openstack-api11:01
*** salv-orlando has quit IRC11:05
*** alex_klimov has joined #openstack-api11:06
*** lucasagomes is now known as lucas-hungry11:38
*** lucas-hungry is now known as lucasagomes12:35
*** etoews has joined #openstack-api13:17
*** sigmavirus24_awa has joined #openstack-api13:57
*** sigmavirus24_awa is now known as sigmavirus2413:57
*** alex_klimov has quit IRC14:22
*** alex_klimov has joined #openstack-api14:23
*** annegentle has joined #openstack-api14:26
*** jaypipes has joined #openstack-api14:29
*** woodster_ has joined #openstack-api14:58
*** vishwanathj has quit IRC15:20
*** Apoorva has joined #openstack-api15:39
*** pballand has joined #openstack-api15:40
*** annegentle has quit IRC16:11
*** cdent has quit IRC16:18
*** e0ne has quit IRC16:20
*** alex_klimov has quit IRC16:20
*** Apoorva has quit IRC16:27
*** etoews has quit IRC16:28
*** jaypipes has quit IRC16:54
*** jaypipes has joined #openstack-api16:55
*** alex_xu has quit IRC17:21
*** alex_xu has joined #openstack-api17:23
*** Apoorva has joined #openstack-api17:25
*** annegentle has joined #openstack-api17:30
*** annegentle has quit IRC17:35
*** lucasagomes is now known as lucas-dinner17:47
*** fzdarsky has joined #openstack-api17:59
*** alex_xu has quit IRC18:04
*** fzdarsky has quit IRC18:07
*** alex_xu has joined #openstack-api18:09
*** vishwanathj has joined #openstack-api18:21
*** vishwana_ has joined #openstack-api18:25
*** vishwanathj has quit IRC18:25
*** vishwana_ has quit IRC18:27
*** fzdarsky has joined #openstack-api18:57
*** etoews has joined #openstack-api19:00
*** e0ne has joined #openstack-api19:03
*** terrylhowe has joined #openstack-api19:22
*** annegentle has joined #openstack-api19:40
*** fzdarsky has quit IRC19:45
*** Apoorva has quit IRC20:15
*** alex_klimov has joined #openstack-api20:18
*** e0ne has quit IRC20:19
*** jaypipes has quit IRC20:20
*** e0ne has joined #openstack-api20:21
*** e0ne has quit IRC20:21
*** Apoorva has joined #openstack-api20:30
*** salv-orlando has joined #openstack-api20:31
*** jaypipes has joined #openstack-api20:31
*** salv-orl_ has quit IRC20:34
*** fzdarsky has joined #openstack-api20:53
*** jaypipes has quit IRC21:04
*** lucas-dinner has quit IRC21:19
*** vishwanathj has joined #openstack-api21:31
*** lucas-dinner has joined #openstack-api21:33
*** vishwanathj has quit IRC21:36
*** devananda has joined #openstack-api21:48
etoewselmiko: i saw the ping in the x-project meeting21:48
devanandaoops, ww. also - hi :)21:48
elmikohey, wb etoews =)21:48
etoewsi'm around for the next 10 min. before leaving on vacation...again.21:49
devanandawanted to raise a question for folks here re: PATCH implementations across projects21:49
devanandasome follow the RFC (https://tools.ietf.org/html/rfc6902) and some do not21:49
elmikoetoews: remove the -2 on https://review.openstack.org/#/c/183698/21:49
elmikodevananda: sure thing, go for it =)21:49
etoewselmiko: done deal21:49
elmikoetoews: thanks, and i've got some ideas to discuss when you are back and in the swing ;)21:50
etoewsmy apologies for not having the time at catch up21:50
elmikono worries, you gave ample warning21:50
devanandaelmiko: what's the best way to raise that discussion / awareness of it, without me writing a formal spec that I'm (honestly) likely to drop the ball on?21:51
elmikodevananda: so, that's a serious issue of debate methinks21:51
elmikoi actually have brought this up before, how should PATCH be applied21:51
elmikoand i don't think we have a consensus21:52
elmikothere is a strong opinion that PATCH should be able to accept partial resource updates, for example21:52
elmikoryansb, miguelgrinberg ^^21:52
devanandasure - which the RFC supports21:52
*** vishwanathj has joined #openstack-api21:52
devanandai mean, partial resource updates is the whole point of PATCH, and what makes it different than PUT21:52
elmikoand then there is the notion of the JSON-patch21:52
devanandathe question is what format the BODY is21:52
elmikoexactly ;)21:52
elmikoi don't think we have an agreement on that21:53
devanandayea -- and a JSON API should take a JSON BODY document for all requests -- not just PUT, POST -- imo :)21:53
ryansbthe strong opinion was, I recall, "if you don't need lists/sets, then a partial PUT body should be ok"21:53
elmikoyea, that makes a certain amount of sense21:53
ryansb"if you need lists/sets/etc, then JSON patch format is nice21:53
devanandaif some service presents a non-JSON API, then sure, it can do a non-JSON PATCH too, right?21:53
elmikomakes sense21:54
devanandaryansb: simpler is good on one hand, but inconsistency is bad, too. if a service expects JSON for only some HTTP methods, that seems less than ideal to me21:54
elmikofor me, the real question is, does an API only accept the JSON-patch notation, or can it also just accept partial resource formats21:54
elmikothat seemed to be a point of discussion for us21:55
devanandaelmiko: afaik, right now, keystone accepts only partial resource, and ironic accepts only JSON-patch.21:55
devanandafor example21:55
devanandaI haven't looked at other services yet21:55
ryansbelmiko: good question. My opinion is that you should stick with one for the whole API21:55
ryansbso if your service just doesn't have lists, then partial actually should be fine21:55
elmikoryansb: agreed, but maybe we should talk about that in the guidance21:55
*** terrylhowe has left #openstack-api21:55
elmikodevananda: sahara is moving down the partial resource path as well21:56
devanandaelmiko: :-/21:56
elmikoi tend to agree with ryansb though, it should at least be consistent across a project. ie don't accept both21:57
elmikoeach project might have different needs21:57
devanandadefinitely agree on that21:57
elmikooh, also barbican is experimenting with partial resource patch mechanics as well21:57
devanandaalso, why would any project not support lists?21:58
elmikoi think the JSON-patch stuff just seems really intimidating at first glance21:58
devanandaelmiko: heh, yea, well, it's not trivial... but it is much clearer what the user is changing21:58
devanandathere's no way to unset (null'ify) a partial resource without it21:58
devanandafor example21:58
elmikogood point21:59
devanandathe only alternative we found would have been to allow PUT of sub-resources21:59
elmikoand i think ryansb meant support for updating lists, is that accurate?21:59
devanandawhich we agreed would only be more compicated21:59
elmikoyea, that's a really good point to bring up when talking about this. it might be a corner case for some, but still worth talking about.22:00
ryansbelmiko: yeah, among other things22:00
*** vishwanathj has quit IRC22:00
ryansbalso lets you do idempotent delete operations, which is nice22:00
elmikodevananda: i think the best advice we can give on the PATCH stuff will center around keeping consistent within the project api, and knowing about the corner cases.22:01
elmikowe should definitely be mentioning both partial resource and json-patch though22:01
devanandaelmiko: is there an RFC for how to implement partial resource PATCH ?22:01
elmikolike an ietf rfc, not sure22:02
elmikosigmavirus24 is the resident rfc wizard... ^^22:02
ryansbelmiko: devananda I believe the answer is no22:02
sigmavirus24== ryansb22:02
ryansbthe RFC's don't say anything about a partial resource format22:03
devanandaI couldn't find one when I had searched (neither did any other ironic API dev)22:03
sigmavirus24One isn't coming to mind immediately22:03
ryansbthey just don't say "you can't"22:03
devanandawhich is another reason we didn't go that route :)22:03
ryansbor "you SHOULD NOT"22:03
* ryansb asked sigmavirus24 the same question some time ago22:03
devanandayea, but there is no guidance on how to do it22:03
elmikodevananda: i think going with 6902 is the sensible approach22:03
elmikobut, the dark side is easier, quicker, ....22:04
*** vishwanathj has joined #openstack-api22:04
ryansbbegun, the PATCH wars have22:04
miguelgrinbergso I try to avoid PATCH as much as I can for these reasons. I think it only makes sense for resources that have a large representation, where sending the whole thing back and forth consumes excessive bandwidth22:06
sigmavirus24Let's just not use PATCH22:06
miguelgrinbergsigmavirus24: there you go. we agree!22:07
sigmavirus24miguelgrinberg: we always agree, we just never know it22:07
elmikoryansb: lol22:07
elmikothis definitely should be a topic for a wider discussion, maybe we should bring it up on the ML?22:08
ryansbyes, let's22:09
miguelgrinbergelmiko: did we ever do a review of how all the projects use it?22:09
ryansbthere's certainly plenty of ground to cover22:09
elmikomiguelgrinberg: not formally, but informally we already have a few different approaches22:09
*** vishwanathj has quit IRC22:09
elmikokeystone, sahara, barbican, are using partial resource, ironic is using 690222:09
miguelgrinbergis it just the partial resources and the json-patch approaches?22:10
elmikoi believe so22:10
elmikobut i'm probably wrong22:10
miguelgrinbergit would be interesting to know if any project use the sub-resource approach instead of patch22:10
elmikoryansb: i'll write an email tomorrow, unless someone beats me to it22:11
elmikomiguelgrinberg: yea, maybe we can do a little review of this during the discussion22:11
ryansbok. I'm heading off on vacation tomorrow, so I'll read it when I get back22:11
elmikohaha, nice. enjoy ryansb !22:12
elmikoi need to head out for tonight, but we can pick this up on ML22:12
elmikothanks again for bringing it up devananda22:12
miguelgrinbergsounds good22:12
devanandaelmiko: yw - and thanks22:15
*** annegentle has quit IRC22:18
*** alex_klimov has quit IRC22:43
*** fzdarsky has quit IRC22:44
*** sigmavirus24 is now known as sigmavirus24_awa23:07
*** fzdarsky has joined #openstack-api23:34
*** lucas-dinner has quit IRC23:51
*** salv-orlando has quit IRC23:52

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!