| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2018-01-10 | |||
| 23:58:40 | openstackgerrit | Merged openstack/python-openstacksdk master: Updated from global requirements https://review.openstack.org/531974 | |
| 23:59:20 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Remove metric service https://review.openstack.org/532697 | |
| #openstack-sdks - 2018-01-11 | |||
| 00:12:04 | mordred | ignore that ^^ | |
| 00:14:50 | mordred | Qiming: if you have a sec, feel like reviewing https://review.openstack.org/#/c/530769/ ? I think it should be ok from your earlier comments | |
| 00:16:45 | mordred | Qiming: also, I pushed up a patch to heat to shift it from using senlinclient to using SDK https://review.openstack.org/#/c/532680/ and a senlin patch https://review.openstack.org/#/c/532641/ that I'd love your feedback on | |
| 02:58:50 | adriant | is it just me, or query params/filters are busted for the openstackSDK? | |
| 02:59:52 | adriant | I'm trying to list stuff via the SDK and the list commands appear to just take key/values as kwargs, but those aren't actually being used since QueryParameters._transpose() appears to gobble them up... | |
| 03:06:43 | adriant | from what I can tell, this is because 'somewhere' on the resource we need to define what they acceptable query params are... or something? but this isn't actually happening anywhere. | |
| 03:07:52 | adriant | no wait... they are actually being used | |
| 03:07:59 | adriant | just not at all for any of the identity | |
| 03:08:26 | adriant | but then the terrifying thing is that after mangling the query, i get returned EVERYTHING | |
| 03:09:26 | adriant | so if I do: sdk.identity.projects(name="some project") | |
| 03:09:43 | adriant | a list with projects of that name, OR nothing | |
| 03:09:47 | adriant | yet I get everything... | |
| 03:10:13 | adriant | because my query is being mangled and the API then does a call for everything | |
| 03:10:34 | adriant | I think it would be much much safer to no mangle the query, just throw a damn error | |
| 03:10:42 | adriant | not* | |
| 03:11:25 | adriant | if you attempt to query with a key that isn't setup to be queried with, throw an error rather than return to me a list of data that I may act upon thinking it matched my query... | |
| 03:11:27 | adriant | when it didn't. | |
| 04:04:57 | openstackgerrit | Adrian Turjak proposed openstack/python-openstacksdk master: Raise error when supplying invalid query params https://review.openstack.org/532723 | |
| 04:05:49 | openstackgerrit | Adrian Turjak proposed openstack/python-openstacksdk master: Raise error when supplying invalid query params https://review.openstack.org/532723 | |
| 04:08:58 | adriant | mordred, briancurtin: : ^ this technically break functionality, but because you're now requiring explicit definition of valid queries in the Resource2 class, rather than just passing them through without checking like Resource did, you can confirm people are being sensible. | |
| 04:10:24 | adriant | Identity has been moved to Resource2, but no one setup any valid queries for it... so trying to list projects and filtering by name returns all projects... which is an unsafe query because you're mangling the query I supplied since no one configured it as valid in the code. | |
| 04:10:55 | adriant | But... since we are configuring valid queries in the code (a good thing), we can stop people from making stupid mistakes by throwing this kind of error. | |
| 04:11:40 | adriant | Which means we actually need to define the queries. I'll dig through and see if there is a patch for adding filters to identityv3, if not, I'll submit that. | |
| 04:14:52 | adriant | hah, fuck, nvm, my bad. the version of the sdk I was using to test doesn't have it... master does... ignore me | |
| 04:18:14 | adriant | Still worth throwing errors for invalid queries because it's safer, since we know what a valid query is | |
| 08:34:28 | openstackgerrit | Zhaokun Fu proposed openstack/python-openstackclient master: add --password option for openstack server create https://review.openstack.org/490718 | |
| 08:48:58 | Mr_Broken | Hey guys, | |
| 08:48:59 | Mr_Broken | Any chance that shade will be supporting SubnetPool functionality ? ( create, delete , manage ? ) | |
| 11:02:43 | openstackgerrit | Merged openstack/python-openstacksdk master: Add query filters for find_network https://review.openstack.org/476410 | |
| 13:31:37 | openstackgerrit | Dmitry Tantsur proposed openstack/api-wg master: [WIP] Add guideline on exposing microversions in SDKs https://review.openstack.org/532814 | |
| 13:47:21 | openstackgerrit | Stephen Finucane proposed openstack/cliff master: sphinxext: Warn if namespace or command pattern invalid https://review.openstack.org/510913 | |
| 14:02:53 | openstackgerrit | Dmitry Tantsur proposed openstack/api-wg master: Add guideline on exposing microversions in SDKs https://review.openstack.org/532814 | |
| 14:03:02 | dtantsur | cdent, mordred, TheJulia, thought dump complete ^^^ :) | |
| 14:18:49 | mordred | dtantsur: ooh, fun reading for this morning | |
| 14:19:07 | dtantsur | haha, I hope you enjoy it :) I did steal a few things from Shade docs ;) | |
| 14:20:12 | mordred | dtantsur: it's good - and it's a hard topic. I think that's an excellent write up | |
| 14:21:09 | dtantsur | cool! | |
| 14:22:20 | mordred | adriant: in shade we do a hybrid approach, which is that we send query parameters to the service if we know about them, but then also filter results client-side before returning them, in case someone wants to filter by something that's not possible as a server-side filter | |
| 14:24:37 | mordred | adriant: I was actually just thinking the other day about adding that post-filtering to Resource2 ... but I think we'd also want to add a flag to control pushdown behavior | |
| 14:25:59 | mordred | because there are three general things people tend to want "pushdown as many filters to the server as possible and apply any remaining filters client-side" "pushdown everything and error if some filters can't be handled server-side" "pushdown nothing and do all filtering client-side" | |
| 17:22:30 | cdent | elmiko, edleafe, dtantsur|afk : https://etherpad.openstack.org/p/api-sig-newsletter please and thank you | |
| 17:24:25 | edleafe | looking... | |
| 17:25:36 | edleafe | looks like cdent has his thesaurus at the ready :) | |
| 17:26:18 | cdent | I am a thesaurus | |
| 17:28:09 | edleafe | cdent: one small wording change, but otherwise LGTM | |
| 17:28:19 | cdent | yah, seems a sensible change | |
| 17:28:24 | cdent | thanks | |
| 17:29:31 | cdent | shipped | |
| 18:10:17 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Add floating IP qos_policy actions https://review.openstack.org/517845 | |
| 18:36:54 | elmiko | sorry cdent, was otherwise occupied :| | |
| 18:37:05 | cdent | no problem | |
| 19:53:17 | adriant | mordred: as long as we make the defaults sensible, and give the user control over them, I'm game. My problem is that if I give it incorrect params, or even correct params thatthe API accepts but we haven't coded in yet, and the code gobbles them up and doesn't tell me... | |
| 19:54:21 | adriant | I'll be screwed when i realise the query I got back includes a list of everything, and none of the filters were actually passed along. | |
| 19:55:00 | adriant | throwing an error at least tells me to submit a patch to add missing filters to the SDK, or that I made a typo and threw in the wrong filter. | |
| 19:55:14 | adriant | silent failure = pain | |
| 19:57:09 | adriant | like I was using 0.9.19 which switched identity to resource2 but didn't yet have the identity filters merged, so when trying to filter project list by name I got back a full list, which... confused me since resource1 just passed the whole dict through. | |
| 19:59:42 | adriant | l" if we don't want to add too much complexity. | |
| 19:59:42 | adriant | mordred: as new services add filters, people will try using them via the SDK sometimes before we add them there, so we either need to make it clear we don't support them yet (error), or a way to explicitly bypass which filters we configure and just pass the filters through raw. The latter of which really could just be a case of: "do a raw API cal | |
| 20:07:00 | openstackgerrit | Merged openstack/osc-lib master: Fix find() interface when attr is not specified https://review.openstack.org/529934 | |
| 21:03:19 | openstackgerrit | Merged openstack/python-openstackclient master: Add floating IP qos_policy actions https://review.openstack.org/517845 | |
| 21:14:49 | openstackgerrit | Michael Johnson proposed openstack/python-openstacksdk master: Re-enable octavia functional tests https://review.openstack.org/531514 | |
| #openstack-sdks - 2018-01-12 | |||
| 02:41:20 | openstackgerrit | Jake Yip proposed openstack/python-openstackclient master: Fix error with image show when image name is None https://review.openstack.org/529464 | |
| 04:21:31 | openstackgerrit | Adrian Turjak proposed openstack/python-openstacksdk master: Raise error when supplying invalid query params https://review.openstack.org/532723 | |
| 04:39:04 | openstackgerrit | Adrian Turjak proposed openstack/python-openstacksdk master: Raise error when supplying invalid query params https://review.openstack.org/532723 | |
| 04:55:12 | adriant | mordred, briancurtin: I think this now mostly makes sense: https://review.openstack.org/#/c/532723 | |
| 05:39:06 | adriant | also, has anyone yet started migrating object_store to resource2 ? I can potentially do that if needed. | |
| 13:48:56 | briancurtin | adriant: go for it | |
| 15:42:40 | jeremyfreudberg | mordred, I've put up a preliminary patch for Sahara APIv2 stuff in python-saharaclient https://review.openstack.org/#/c/533066/ | |
| 15:43:10 | jeremyfreudberg | in my part time -ness i've only had time to do things the hacky way | |
| 15:43:17 | jeremyfreudberg | but that should make it easier for your to swoop in | |
| 15:44:07 | mordred | jeremyfreudberg: awesome! I'll open it up and look in just a sec! | |
| 15:45:06 | jeremyfreudberg | mordred: appreciate it | |
| 16:50:30 | mordred | adriant: awesome, thanks! patch looks great - and I think it leaves the door open to later adding the fancier thing I was suggesting - your patch makes things safe at least :) | |
| 16:51:01 | mordred | adriant: also, yes, what briancurtin said - would love help migrating object_store - although if you don't get to it in the next day or two it's high up there on my list | |
| 17:20:57 | openstackgerrit | Monty Taylor proposed openstack/os-service-types master: Add factory helper function which returns a singleton https://review.openstack.org/533250 | |
| 18:41:43 | openstackgerrit | Merged openstack/python-openstacksdk master: Prefer links dicts for pagination https://review.openstack.org/530769 | |
| 18:44:58 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Rationalize logging helpers and docs https://review.openstack.org/532634 | |
| 18:44:58 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Merge user and users sections of the docs https://review.openstack.org/532633 | |
| 18:44:59 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Remove meter service https://review.openstack.org/533264 | |
| 18:44:59 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Add ability to register non-official services https://review.openstack.org/533263 | |
| 18:45:00 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Remove metric service https://review.openstack.org/533265 | |
| 21:42:00 | openstackgerrit | Merged openstack/python-openstacksdk master: Add reno for tag support on heat stacks https://review.openstack.org/527381 | |
| 23:29:21 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Merge user and users sections of the docs https://review.openstack.org/532633 | |
| 23:29:22 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Add ability to register non-official services https://review.openstack.org/533263 | |
| 23:29:22 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Rationalize logging helpers and docs https://review.openstack.org/532634 | |
| 23:29:23 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Remove metric service https://review.openstack.org/533265 | |
| 23:29:23 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Remove meter service https://review.openstack.org/533264 | |
| 23:29:24 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Migrate object_store to resource2/proxy2 https://review.openstack.org/533323 | |
| 23:29:24 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Remove message v1 support https://review.openstack.org/533322 | |
| 23:30:34 | mordred | adriant: I started on object_store ^^ | |
| 23:31:44 | mordred | adriant: I didn't get a chance to touch tests at all - but I figured I'd push it up in case you felt like hacking - I'm about out for the evening, feel free to take that patch over | |
| 23:31:54 | mordred | adriant: or to ignore it | |
| 23:38:21 | openstackgerrit | Monty Taylor proposed openstack/python-openstacksdk master: Base devstack sanity check on os_service_types https://review.openstack.org/533325 | |
| 23:38:54 | mordred | johnsom: ^^ did a followup to your patch taking a stab at using ost for that devstack test - no clue if that'll work :) | |
| 23:39:48 | johnsom | Cool! | |
| #openstack-sdks - 2018-01-13 | |||
| 01:14:05 | openstackgerrit | Merged openstack/python-openstacksdk master: Re-enable octavia functional tests https://review.openstack.org/531514 | |
| 05:36:03 | openstackgerrit | Merged openstack/python-openstacksdk master: Remove legacy client factory functions https://review.openstack.org/532347 | |
| 06:15:43 | adriant | mordred: will take a look at it monday my time when I'm back at work. I can try it out, potentially fix some tests, or at least leave some review. | |