| *** annegentle has joined #openstack-api | 00:06 | |
| *** annegentle has quit IRC | 00:31 | |
| *** annegentle has joined #openstack-api | 00:38 | |
| *** annegentle has quit IRC | 00:48 | |
| *** annegentle has joined #openstack-api | 00:49 | |
| *** annegentle has quit IRC | 01:08 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 01:19 | |
| *** kaufer has joined #openstack-api | 02:52 | |
| *** terrylhowe has quit IRC | 03:17 | |
| *** terrylhowe has joined #openstack-api | 03:37 | |
| *** kaufer has quit IRC | 04:10 | |
| *** terrylhowe has quit IRC | 06:38 | |
| *** etoews has quit IRC | 07:39 | |
| *** e0ne has joined #openstack-api | 09:56 | |
| *** e0ne is now known as e0ne_ | 09:56 | |
| *** e0ne_ has quit IRC | 10:05 | |
| *** e0ne has joined #openstack-api | 10:10 | |
| *** terrylhowe has joined #openstack-api | 11:07 | |
| *** cdent has joined #openstack-api | 12:03 | |
| *** e0ne is now known as e0ne_ | 12:35 | |
| *** e0ne_ has quit IRC | 12:45 | |
| *** kaufer has joined #openstack-api | 12:47 | |
| *** annegentle has joined #openstack-api | 13:37 | |
| *** e0ne has joined #openstack-api | 13:44 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 13:51 | |
| *** etoews_ has joined #openstack-api | 14:35 | |
| etoews_ | just realize the meeting is an hour later for me because dst. | 14:55 |
|---|---|---|
| etoews_ | s/ze/zed/ | 14:55 |
| *** etoews_ is now known as etoews | 14:55 | |
| annegentle | oh! | 15:00 |
| *** e0ne is now known as e0ne_ | 15:01 | |
| annegentle | so is it in an hour from now? | 15:01 |
| *** katco has joined #openstack-api | 15:02 | |
| annegentle | etoews: so it's at 11 Central? | 15:02 |
| etoews | yep. http://www.worldtimebuddy.com/?qm=1&lid=100,4671654&h=100&date=2015-3-12&sln=16-17 | 15:04 |
| etoews | annegentle: ^ | 15:04 |
| sigmavirus24 | heh | 15:05 |
| sigmavirus24 | It's so disorienting | 15:05 |
| sigmavirus24 | but this is why I have calendar meetings set up in utc | 15:05 |
| * katco detests dst | 15:05 | |
| *** e0ne_ has quit IRC | 15:06 | |
| etoews | sigmavirus24: i just updated the meeting agenda. please double check to make sure i didn't clobber anything still relevant from the last meeting. | 15:12 |
| sigmavirus24 | will do etoews | 15:12 |
| sigmavirus24 | MY TOPIC HAS FAILED ME | 15:12 |
| sigmavirus24 | The meeting agenda link should be there too | 15:13 |
| sigmavirus24 | =P | 15:13 |
| etoews | topic needs more links | 15:14 |
| sigmavirus24 | yep | 15:19 |
| sigmavirus24 | Also, versioning: I never got around to writing the spec for that again | 15:19 |
| *** e0ne has joined #openstack-api | 15:21 | |
| *** rosmaita has joined #openstack-api | 15:24 | |
| -sigmavirus24- Meeting in 14 minutes in #openstack-meeting-3 | 15:47 | |
| *** e0ne is now known as e0ne_ | 16:13 | |
| *** e0ne_ is now known as e0ne | 16:17 | |
| *** TravT has joined #openstack-api | 17:00 | |
| *** lakshmiS has joined #openstack-api | 17:00 | |
| *** wko has joined #openstack-api | 17:00 | |
| *** smc7 has joined #openstack-api | 17:00 | |
| sigmavirus24 | Woo more people | 17:00 |
| rosmaita | etoews: this channel is logged, right? so if we continue, it will be logged for posterity? | 17:01 |
| sigmavirus24 | rosmaita: yes | 17:01 |
| rosmaita | cool | 17:01 |
| etoews | if the conversation is continuing here from the api_wg meeting here's the history. http://eavesdrop.openstack.org/meetings/api_wg/2015/api_wg.2015-03-12-16.00.log.html | 17:01 |
| sigmavirus24 | also the conclusion should be brought to the ML anyway | 17:01 |
| cdent | we should be versioning media types not endpoints | 17:01 |
| sigmavirus24 | cdent++ | 17:01 |
| rosmaita | ok, so as far as advice for our particular case ... | 17:01 |
| rosmaita | so far, we've got: | 17:01 |
| rosmaita | ok to be experimental ... but mark it clearly in the docs as such | 17:02 |
| rosmaita | and state more clearly what experimental means | 17:02 |
| elmiko | cdent: not sure i follow, could you elaborate about versioing media types? | 17:02 |
| rosmaita | experimental == a breaking change may be introduced | 17:03 |
| cdent | github's api is one place I've seen it done, elmiko | 17:03 |
| elmiko | cdent: ack, i'll take a look at their docs | 17:03 |
| cdent | basically you make a request with an Accept header that declares the media type you want the reqponse to be in: application/vnd.github+json or some such (I can't remember) | 17:03 |
| rosmaita | to be consistent with current openstack though, we will need to include a version in the url? like /x1/whatever | 17:04 |
| cdent | and the version is built into the media type, sometimes in the vnd.github part, sometime as a param | 17:04 |
| elmiko | interesting | 17:04 |
| * cdent is in another meeting now, so has partial attentio | 17:04 | |
| cdent | n | 17:04 |
| cdent | Generally speaking I think we should be using content negotiation _far_ more. | 17:05 |
| stevelle | cdent: that does work pretty well, and covers most of what is being attempted with API microversions, but some changes still wont fit. | 17:05 |
| cdent | it's no panacea, that's for sure | 17:05 |
| stevelle | the microversions solution we are seeing in OpenStack could largely be covered through that though | 17:05 |
| etoews | stevelle: are you saying there is content negotiation in nova microversions? | 17:07 |
| etoews | i don't see anything related to that here http://specs.openstack.org/openstack/nova-specs/specs/kilo/approved/api-microversions.html | 17:08 |
| etoews | am i missing something? | 17:08 |
| stevelle | etoews: not nova specifically | 17:08 |
| stevelle | I'm less familiar with that one particularly | 17:08 |
| etoews | stevelle: oh were you talking about an industry attempt at API microversions? | 17:09 |
| *** wko has left #openstack-api | 17:09 | |
| TravT | So for new endpoints, the /x1/ or /x/ in the url still seemed like a nice way to at least see that in the registry. | 17:09 |
| stevelle | I'm making my point poorly while distracted by another conversation etoews, sorry | 17:10 |
| etoews | stevelle: no worries | 17:10 |
| *** kaufer has quit IRC | 17:10 | |
| etoews | TravT: that would certainly elevated the visibility but requiring a version in some header would too. | 17:11 |
| etoews | in that you reject the request if it does have the version in the header | 17:11 |
| etoews | and the error message says, put x1 in the such and such header. by using this header you acknowledge that you are using an experimental api! | 17:12 |
| miguelgrinberg | cdent: the problems I see with versioning media types are (1) not user friendly, since you have to use these weird content-types (2) promotes the use of a single endpoint handler function for all versions, which leads to harder to maintain code | 17:12 |
| etoews | in fact, that might be a better way to do it because you can actually get a human readable message to the user | 17:12 |
| etoews | i'm just brainstorming here though | 17:13 |
| stevelle | etoews: with regards to additive changes to resources that can be dealt with through content negotiation instead. I am still a believer in major version numbers but I'm in a minority there I suspect | 17:14 |
| etoews | ah | 17:18 |
| etoews | well...i've got to head out. thanks all. | 17:18 |
| elmiko | later | 17:19 |
| *** smc7 has left #openstack-api | 17:20 | |
| cdent | miguelgrinberg: I agree there are challenges, especially with pre-existing apis, howver I reckon 1 could be overcome with a reasonable default (if you send application/json instead of the special media type you get the default) and 2 is made harder by the mostly dreadful frameworks that exist | 17:21 |
| cdent | (dreadful in the sense that the routing and dispatch models don't map well) | 17:22 |
| *** kaufer has joined #openstack-api | 17:23 | |
| miguelgrinberg | cdent: and the problem with allowing the client to not specify the version to get the latest, is that when you bump the version that client might break | 17:23 |
| miguelgrinberg | it's a hard problem to solve, I don't think there is a satisfactory solution | 17:23 |
| * cdent likes to think of things breaking as a way of ensuring that people pay attention ;) | 17:24 | |
| elmiko | hehe | 17:24 |
| miguelgrinberg | I personally prefer versions in URLs, accompanied by careful design to try to minimize the need to require new versions except for truly major changes | 17:25 |
| cdent | I think you're probably right that that is the far more pragmatic way | 17:26 |
| stevelle | I'm mostly with miguelgrinberg on this, but I don't expect to see that working as well in this context. OpenStack changes more than any set of APIs I can compare to. | 17:33 |
| miguelgrinberg | stevelle: unfortunate, but true | 17:34 |
| stevelle | and a lot of the getting-there-from-here challenges for the existing APIs makes this particularly difficult | 17:34 |
| *** e0ne has quit IRC | 17:37 | |
| *** cdent has quit IRC | 17:58 | |
| *** annegentle has quit IRC | 18:00 | |
| *** lakshmiS has quit IRC | 18:00 | |
| *** annegentle has joined #openstack-api | 18:12 | |
| *** annegentle has quit IRC | 18:15 | |
| *** annegentle has joined #openstack-api | 18:15 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 19:27 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 19:28 | |
| *** e0ne has joined #openstack-api | 19:39 | |
| *** openstackgerrit has quit IRC | 19:40 | |
| *** openstackgerrit has joined #openstack-api | 19:40 | |
| *** e0ne has quit IRC | 20:30 | |
| *** e0ne has joined #openstack-api | 20:34 | |
| *** kaufer has quit IRC | 21:42 | |
| *** annegentle has quit IRC | 21:56 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:02 | |
| *** e0ne has quit IRC | 22:05 | |
| etoews | rosmaita: TravT: did you notice the "experimental" tag (for lack of a better term) in nova microversions? http://specs.openstack.org/openstack/nova-specs/specs/kilo/approved/api-microversions.html#client-interaction | 22:20 |
| etoews | X-OpenStack-Nova-API-Version: version_number, experimental | 22:21 |
| etoews | The first header specifies the version number of the API which was executed. Experimental is only returned if the operator has made a modification to the API behaviour that is non standard. This is only intended to be a transitional mechanism while some functionality used by cloud operators is upstreamed and it will be removed within a small number of | 22:21 |
| etoews | releases. | 22:21 |
| etoews | i'm not sure if that would work with your requirements or not | 22:22 |
| TravT | etoews: that actually really interesting. I'm down in the weeds on something at the moment. You might want to reply to the [openstack-dev] [Glance] Experimental API thread with that info | 22:23 |
| rosmaita | etoews: had not seen that, thanks for the ref | 22:23 |
| *** annegentle has joined #openstack-api | 22:56 | |
| *** annegentle has quit IRC | 23:02 | |
| *** annegentle has joined #openstack-api | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!