| *** stevelle has joined #openstack-sdks | 00:04 | |
| *** Qiming has joined #openstack-sdks | 00:05 | |
| *** Qiming has quit IRC | 00:18 | |
| *** etoews has quit IRC | 01:09 | |
| *** mattfarina has joined #openstack-sdks | 01:13 | |
| *** mattfarina has quit IRC | 01:16 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 01:18 | |
| *** Qiming has joined #openstack-sdks | 01:41 | |
| *** Qiming_ has joined #openstack-sdks | 01:42 | |
| *** Qiming has quit IRC | 01:46 | |
| *** Qiming__ has joined #openstack-sdks | 01:47 | |
| *** Qiming_ has quit IRC | 01:50 | |
| *** Yanyanhu has joined #openstack-sdks | 01:54 | |
| *** Qiming__ is now known as Qiming | 01:58 | |
| *** pm90_ has joined #openstack-sdks | 04:33 | |
| *** pm90__ has joined #openstack-sdks | 04:46 | |
| *** pm90_ has quit IRC | 04:47 | |
| *** pm90__ has quit IRC | 05:04 | |
| *** toabctl has quit IRC | 05:21 | |
| *** jlvillal has quit IRC | 05:26 | |
| *** jlvillal has joined #openstack-sdks | 07:29 | |
| *** karimb has joined #openstack-sdks | 07:37 | |
| *** aufi has joined #openstack-sdks | 07:42 | |
| *** bnemec has quit IRC | 09:13 | |
| *** bnemec has joined #openstack-sdks | 09:13 | |
| *** trown|outttypeww is now known as trown | 09:27 | |
| *** Yanyanhu has quit IRC | 10:15 | |
| *** Qiming has quit IRC | 10:27 | |
| *** jamielennox is now known as jamielennox|away | 10:58 | |
| *** chlong has joined #openstack-sdks | 11:48 | |
| *** thrash|g0ne is now known as thrash | 11:50 | |
| *** stevemar has joined #openstack-sdks | 12:38 | |
| *** bknudson has quit IRC | 12:42 | |
| *** bknudson has joined #openstack-sdks | 13:41 | |
| *** britthouser has joined #openstack-sdks | 13:56 | |
| *** bradbeam_ has joined #openstack-sdks | 14:00 | |
| *** britthou_ has joined #openstack-sdks | 14:01 | |
| Shrews | mordred: so, quick off-the-top-of-my-head example of how this new o-c-c command line might operate: http://paste.openstack.org/show/222839/ | 14:01 |
|---|---|---|
| Shrews | good? bad? missing stuff? | 14:01 |
| *** bradbeam_ has left #openstack-sdks | 14:01 | |
| *** bradbeam has quit IRC | 14:02 | |
| Shrews | or maybe you or greghaynes has already started on something | 14:03 |
| mordred | Shrews: I think that looks great. dtroyer was saying he thinks the commands should actually be added to osc | 14:03 |
| Shrews | oh, hrm | 14:03 |
| Shrews | well, that could work too | 14:03 |
| mordred | so like "openstack cloud add" and "openstack cloud merge" maybe? (with the same semantics you have there) | 14:04 |
| *** britthouser has quit IRC | 14:04 | |
| Shrews | i'm cool with that | 14:04 |
| mordred | whee! | 14:04 |
| dtroyer | it would be simple to do as a plugin, could me merged into the repo too | 14:04 |
| dtroyer | I'll likely be an ass about the command structure though… ;) | 14:04 |
| mordred | Shrews, dtroyer: I was also thinking a "openstack cloud use" might be nice ... let you set a 'current cloud' on the command line | 14:05 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:05 | |
| dtroyer | ah, I did think of that wrt interactive use, wasn't sure about from the shell though | 14:05 |
| mordred | dtroyer: well, I've never written an osc command and I don't think Shrews has either - so it'll be fun | 14:05 |
| Shrews | i've never even seen osc code, so... yeah | 14:06 |
| mordred | dtroyer: I was thinking from the shell maybe openstack cloud use would write out a ~/.openstackcloud or something with the cloud name in it or something | 14:06 |
| mordred | dtroyer: I have not fully thought this through :) | 14:06 |
| mordred | dtroyer: just wait till I bother you with wanting "openstack server list" to be able to handle multi-clouds :) | 14:07 |
| dtroyer | my hesitation on doing it interactively is in users forgetting and wondering why it Doesn't Work(TM). | 14:07 |
| mordred | dtroyer: yah. that's the biggest issue there | 14:07 |
| dtroyer | mordred: heh, I started that one already! | 14:07 |
| mordred | dtroyer: NEAT! | 14:07 |
| mordred | it's like we share one brain between us | 14:07 |
| dtroyer | that's where the damn cats came from! | 14:08 |
| Shrews | mordred: you having half a brain sounds about right | 14:08 |
| * Shrews hides | 14:08 | |
| * mordred throws a thousand used loofas at Shrews | 14:08 | |
| *** mattfarina has joined #openstack-sdks | 14:10 | |
| mordred | greghaynes: ^^ when you awaken, you probably want to read that | 14:17 |
| dtroyer | mordred, Shrews: I did a _really_ _quick_ rename of the sample OSC plugin to osc-cloud if you want to look at what an OSC plugin looks like: https://github.com/dtroyer/osc-cloud | 14:18 |
| mordred | dtroyer: do you thnk it should be a plugin and not just in osc? | 14:21 |
| * mordred could go either way, honestly | 14:21 | |
| mordred | since osc depends on o-c-c - if occ provided an osc plugin, it would be six in one half a dozen in the other | 14:21 |
| dtroyer | I'd start as a plugin, we could merge it later? I just start experimantal thigs as plugins these days | 14:21 |
| dtroyer | but I wouldn't reject adding it directly either | 14:22 |
| *** britthou_ has quit IRC | 14:22 | |
| mordred | dtroyer: (you didn't git add osccloud | 14:22 |
| dtroyer | I said it was a quick rename of the default. didn't change the commands yet either ;) | 14:22 |
| mordred | dtroyer: there's no source dir though ... | 14:22 |
| mordred | unless you don't need any source code to write pugins. *MIND*BLOWN* | 14:23 |
| dtroyer | I wish I could claim lack of caffiene…there it is | 14:23 |
| mordred | :) | 14:25 |
| dtroyer | making this into a cookiecutter thing is on my list…someday… | 14:25 |
| Shrews | dtroyer: mordred: i assume a plugin wouldn't require a blueprint spec to first get approved, yeah? | 14:27 |
| mordred | uhm. no | 14:27 |
| Shrews | b/c, i'm all about simplicity | 14:27 |
| mordred | Shrews: so - looking at that, it seems like if we add the needed source to occ - but _don't_ add cliff or openstackclient to requirements - instead, add them to test-requirements | 14:27 |
| mordred | that way we can functional test that the occ plugin works with openstackclient | 14:28 |
| mordred | but if someone just installs occ without osc, they don't get osc depends sucked in | 14:28 |
| mordred | but we can register the entry point | 14:28 |
| mordred | so if they do later install osc, they will magically get everythign working | 14:28 |
| mordred | or? | 14:28 |
| Shrews | that magic sounds acceptable | 14:29 |
| dtroyer | I wouldn't put the OSC (cliff) command classes into o-c-c, just the implementation underneath them | 14:29 |
| mordred | dtroyer: you'd make a totally separate repo for the plugin? | 14:29 |
| *** shakamunyi has joined #openstack-sdks | 14:29 | |
| dtroyer | if this were optional, yes, the more I think about it, in the end here, no | 14:30 |
| dtroyer | if o-c-c in OSC were optional | 14:30 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in object_store proxy https://review.openstack.org/183063 | 14:30 |
| mordred | yah. seems like something that just wants to be a thing | 14:30 |
| dtroyer | so that plugin is good for a pattern to build the command classes, less to look at than OSC itself | 14:30 |
| dtroyer | and that pushes me to putting the commands into OSC repo | 14:31 |
| mordred | I believe I am now of that opinion too, from the talking | 14:31 |
| dtroyer | so given that, it deserves a new top-level dir: openstackclient.cloud or something | 14:32 |
| dtroyer | I've been putting litle stuff like this into openstackclient.common but that feels dirty | 14:33 |
| * mordred likes dirty things | 14:33 | |
| stevemar | keep things pg mordred ! | 14:33 |
| dtroyer | that's where the module andtiming commands like for example | 14:34 |
| Shrews | so... no plugin? | 14:34 |
| * mordred hands stevemar something inappropriate | 14:34 | |
| stevemar | eep! | 14:34 |
| mordred | Shrews: yah. it sounds like just a patch to osc | 14:34 |
| stevemar | dtroyer, the common module has been getting a bit overloaded | 14:34 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in image proxies https://review.openstack.org/182804 | 14:35 |
| dtroyer | stevemar: I think it stll makes sense for commands that cross REST APIs, but for internal bits like this I'd favor a better location | 14:35 |
| stevemar | ++ | 14:35 |
| stevemar | quota/limit/extensions can stay there | 14:36 |
| stevemar | do we have utils in there? | 14:36 |
| stevemar | i suppose they could stay put | 14:36 |
| *** toabctl has joined #openstack-sdks | 14:37 | |
| *** aufi has quit IRC | 14:37 | |
| *** Shrews has quit IRC | 14:38 | |
| *** mordred has quit IRC | 14:40 | |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Orchestration proxy changes https://review.openstack.org/183066 | 14:41 |
| *** mordred has joined #openstack-sdks | 14:42 | |
| *** Shrews has joined #openstack-sdks | 14:43 | |
| *** barra204 has joined #openstack-sdks | 14:43 | |
| Shrews | ugh | 14:46 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in telemetry proxy https://review.openstack.org/183068 | 14:47 |
| dtroyer | stevemar: what do you think about skipping osc meeting today? I didn't do a release and probably will not this close to the weekend | 14:50 |
| stevemar | dtroyer, thats fine with me, my schedule is packed til the summit | 14:52 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in volume proxy https://review.openstack.org/183073 | 14:54 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Orchestration proxy changes https://review.openstack.org/183066 | 14:56 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in object_store proxy https://review.openstack.org/183063 | 14:57 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Common list method for proxies https://review.openstack.org/178302 | 15:10 |
| *** pm90_ has joined #openstack-sdks | 15:17 | |
| * greghaynes awakens to a great many highlights | 15:19 | |
| greghaynes | mordred: ohai | 15:19 |
| mordred | greghaynes: tl;dr - we're going to add your idea to osc | 15:19 |
| mordred | greghaynes: Shrews may alraedy be working on it | 15:19 |
| greghaynes | ok, trying to parse the bits about it being optional and living in multiple places? | 15:19 |
| Shrews | well, not *yet* | 15:20 |
| greghaynes | does osc not currently depend on os-c-c} | 15:20 |
| *** pm90_ has quit IRC | 15:20 | |
| dtroyer | greghaynes: it does…skip the commants about OSC plugins, we're not doing that part | 15:20 |
| *** pm90_ has joined #openstack-sdks | 15:20 | |
| greghaynes | heh | 15:21 |
| greghaynes | ok, awesome | 15:21 |
| greghaynes | and command structure, I was thinking <cmd> cloud add <cloudname> <destfile? | 15:21 |
| Shrews | greghaynes: is this something you want to work on? i wasn't sure of your interest level... was just going to start doing it if no one else was | 15:23 |
| greghaynes | Shrews: Yea, I can do it, started poking a bit at it in o-c-c yesterday | 15:23 |
| greghaynes | hah | 15:24 |
| Shrews | greghaynes: cool. i can stick to shade & ansible things, then | 15:24 |
| greghaynes | oh, mordred did a hack way | 15:24 |
| greghaynes | nice | 15:24 |
| mordred | I did? | 15:24 |
| mordred | oh - the just-add-it-to-devstack thing? | 15:24 |
| greghaynes | yep | 15:24 |
| mordred | yah | 15:24 |
| mordred | I think once there is an osc command, we should just run that | 15:24 |
| Shrews | yeah | 15:26 |
| briancurtin | terrylhowe: i am going to just disable any tests that use httpretty because it is somehow once again fucking this entire process up even after hacking in that version. once the list change goes through they won't actually be needed. | 15:31 |
| *** etoews has joined #openstack-sdks | 15:32 | |
| terrylhowe | briancurtin: sounds good | 15:34 |
| stevemar | dtroyer, so the meeting today is cacled? | 15:37 |
| stevemar | hehe | 15:37 |
| * stevemar fights against the urge to reply to the ML with a joke | 15:37 | |
| terrylhowe | no meeting today? | 15:39 |
| dtroyer | terrylhowe: no | 15:40 |
| dtroyer | stevemar: that's evidence enough on why it should be... | 15:41 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Remove httpretty https://review.openstack.org/183094 | 15:42 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Remove httpretty https://review.openstack.org/183094 | 15:42 |
| *** etoews_ has joined #openstack-sdks | 15:47 | |
| *** etoews_ has quit IRC | 15:48 | |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for get calls in object_store proxy https://review.openstack.org/183063 | 15:56 |
| mordred | dtroyer: actually - you like go, right? | 16:04 |
| mordred | or someone in here does iirc | 16:04 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Remove httpretty https://review.openstack.org/183094 | 16:05 |
| greghaynes | surely someone on the internet likes go | 16:05 |
| dtroyer | GO GO GO!!! | 16:07 |
| dtroyer | but I grew up in Pascal so it's like going home | 16:07 |
| dtroyer | (pun intended) | 16:07 |
| sigmavirus24 | greghaynes: not for long? | 16:12 |
| greghaynes | hah | 16:12 |
| sigmavirus24 | <trolling>the only good part of go is the package management</trolling> | 16:14 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for get calls in network proxy https://review.openstack.org/182841 | 16:22 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for get calls in telemetry proxy https://review.openstack.org/183068 | 16:26 |
| *** shakamunyi has quit IRC | 16:27 | |
| *** barra204 has quit IRC | 16:28 | |
| openstackgerrit | Merged stackforge/python-openstacksdk: Remove httpretty https://review.openstack.org/183094 | 16:36 |
| briancurtin | terrylhowe: any thought on https://review.openstack.org/#/c/178302/ for list calls? i'm trying to squeeze in a change for head as well, but should be pretty much in line with what we've already done for get since it's similar | 16:39 |
| terrylhowe | briancurtin: I’m trying to slog through all these get changes and then list | 16:40 |
| briancurtin | ok cool | 16:40 |
| terrylhowe | I should get to list soon | 16:40 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for get calls in volume proxy https://review.openstack.org/183073 | 16:43 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Orchestration proxy changes https://review.openstack.org/183066 | 16:44 |
| briancurtin | terrylhowe: yeah, documenting that stuff is going to be really hard, but the one that's in there appears to be pretty accurate - should i still remove that one or just not add any others while finishing it out? I had looked at making query params some type of list attached to each resource, and maybe we can do that later on | 17:04 |
| terrylhowe | well, not add any others will speed things up for sure. | 17:04 |
| terrylhowe | query parms on the resource is an interesting idea | 17:05 |
| briancurtin | terrylhowe: i had looked at it being a list or a dict, and then some type of cross-check against what params are sent to the list method. might work, might be too much. we kind of do need the prop functionality there, as evidenced by the changes-since/changes_since one in that review | 17:07 |
| briancurtin | terrylhowe: and yeah, i'll stick with not adding any others. the follow-on changes would purely be accepting the **params and setting the appropriate paginated= kwarg | 17:31 |
| terrylhowe | sounds good briancurtin I’ll post a couple other discussion points, but I still want to mess with it a little | 17:37 |
| *** jsavak has joined #openstack-sdks | 17:51 | |
| greghaynes | mordred: annoying thing about this outputting clouds.yaml - if we grab the config to output from the os-c-c config that osc loads then were going to be serializing all the defaults into the clouds.yaml | 18:06 |
| mordred | briancurtin: I was about to ask how you set up your test_cloud cloud for your functional tests ... but then realized that you arent' running those in the gate yet | 18:06 |
| mordred | greghaynes: I'm not following how that's annoying? | 18:06 |
| greghaynes | hrm | 18:07 |
| mordred | greghaynes: but it's possible a different sequence of words would help my age-adled brain | 18:07 |
| greghaynes | no, its a valid point | 18:07 |
| greghaynes | well see how it works because whynot | 18:08 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Common list method for proxies https://review.openstack.org/178302 | 18:25 |
| *** jlvillal has quit IRC | 18:36 | |
| *** jlvillal has joined #openstack-sdks | 18:41 | |
| *** pm90_ has quit IRC | 19:08 | |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Common head method for proxies https://review.openstack.org/183183 | 19:15 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Common list method for proxies https://review.openstack.org/178302 | 19:24 |
| *** stevemar has quit IRC | 19:25 | |
| *** pm90_ has joined #openstack-sdks | 19:45 | |
| *** openstackgerrit has quit IRC | 19:52 | |
| *** openstackgerrit has joined #openstack-sdks | 19:52 | |
| *** trown is now known as trown|outttypeww | 20:01 | |
| openstackgerrit | Terry Howe proposed stackforge/python-openstacksdk: Common head method for proxies https://review.openstack.org/183183 | 20:42 |
| briancurtin | terrylhowe: thanks for getting that head one | 20:43 |
| terrylhowe | np | 20:43 |
| briancurtin | going through the list changes now | 20:43 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Common head method for proxies https://review.openstack.org/183183 | 20:48 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for compute proxy list calls https://review.openstack.org/183206 | 20:53 |
| openstackgerrit | Terry Howe proposed stackforge/python-openstacksdk: Fix comment error on gets and heads https://review.openstack.org/183207 | 20:54 |
| terrylhowe | briancurtin: ^^ might want to take a quick look at that one and rebase if it looks good | 20:55 |
| terrylhowe | I’d approved probably 4 get changes before I noticed it | 20:56 |
| briancurtin | terrylhowe: ah, crap. yeah. am currently making a minor tweak to _get and will rebase the rest of those | 20:57 |
| *** chlong has quit IRC | 21:08 | |
| *** jsavak has quit IRC | 21:10 | |
| *** openstackgerrit has quit IRC | 21:22 | |
| *** openstackgerrit has joined #openstack-sdks | 21:22 | |
| *** mattfarina has quit IRC | 21:44 | |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Move compute limits to new _get API https://review.openstack.org/183322 | 21:50 |
| openstackgerrit | greghaynes proposed openstack/os-client-config: Add set_one_cloud method https://review.openstack.org/183323 | 21:50 |
| openstackgerrit | greghaynes proposed openstack/python-openstackclient: Create cloud update command https://review.openstack.org/183324 | 21:51 |
| greghaynes | mordred: ^ | 21:51 |
| mordred | greghaynes: NEAT! | 21:53 |
| greghaynes | I kind of need to figure out the tests for the osc change | 21:54 |
| greghaynes | but seems to work | 21:54 |
| greghaynes | SpamapS: Shrews ^ youall too | 22:00 |
| greghaynes | might find that of interest | 22:00 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Move compute limits to new _get API https://review.openstack.org/183322 | 22:10 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Fix comment error on gets and heads https://review.openstack.org/183207 | 22:18 |
| *** bknudson has quit IRC | 22:20 | |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for compute proxy list calls https://review.openstack.org/183206 | 22:24 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in image proxies https://review.openstack.org/182804 | 22:25 |
| mordred | greghaynes: left a -1 on the o-c-c change | 22:26 |
| mordred | greghaynes: it's a nit-ish thing - but easy enough tofix | 22:26 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for database proxy gets https://review.openstack.org/182788 | 22:27 |
| openstackgerrit | Brian Curtin proposed stackforge/python-openstacksdk: Changes for get calls in image proxies https://review.openstack.org/182804 | 22:28 |
| openstackgerrit | Amey Bhide proposed openstack/python-openstackclient: Add support for v2 image update commands https://review.openstack.org/182504 | 22:42 |
| *** shakamunyi has joined #openstack-sdks | 22:47 | |
| *** barra204 has joined #openstack-sdks | 22:47 | |
| mordred | jamielennox|away: so, now that keystoneauth exists - porting openstacksdk to it should be a thing, right? | 23:02 |
| sigmavirus24 | mordred: but first, is there documentation? | 23:02 |
| dtroyer | please no? sdk should have _no_ dependencies like that or we'll never leave the version hell we are in | 23:03 |
| mordred | sigmavirus24: not yet | 23:03 |
| mordred | dtroyer: I think keystoneauth is the one you do want | 23:03 |
| mordred | because it doesnt' depend on the other things | 23:03 |
| briancurtin | if it's not documented it won't be considered at all. | 23:03 |
| mordred | well, it _will_ have by the 1.0 release | 23:03 |
| *** barra204 has quit IRC | 23:04 | |
| briancurtin | already had enough of a time dealing with current "documentation" | 23:04 |
| *** shakamunyi has quit IRC | 23:04 | |
| *** redrobot has quit IRC | 23:04 | |
| dtroyer | but the server-side vs client-side differences will not go away in that case | 23:04 |
| mordred | dtroyer: what do you mean? | 23:04 |
| dtroyer | that's why the SDK is client-saviour for non-server use | 23:04 |
| mordred | right | 23:04 |
| mordred | keystoneauth is just about getting a session to auth against a keystone | 23:05 |
| mordred | it's not all the server related things | 23:05 |
| *** redrobot has joined #openstack-sdks | 23:05 | |
| mordred | unless I am missing something | 23:05 |
| dtroyer | but if both server and client use it, then we never leave the dependency version hell | 23:05 |
| *** redrobot is now known as Guest19895 | 23:05 | |
| mordred | aroo? | 23:05 |
| mordred | auth-ing against a keystone shold be versionless? | 23:06 |
| * mordred may not grok the problem | 23:06 | |
| dtroyer | the library itself. installing master clients on a stable/xyz server is broken | 23:06 |
| dtroyer | why we did stable/kilo for client and libs | 23:06 |
| mordred | nod | 23:06 |
| mordred | one sec | 23:06 |
| *** morganfainberg has joined #openstack-sdks | 23:07 | |
| morganfainberg | mordred: o/ | 23:07 |
| mordred | if openstacksdk can't use it without hell, then something is broken - but I've asked morgan to come in because he knows plans better | 23:07 |
| mordred | morganfainberg: we're talking about keystoneauth | 23:08 |
| mordred | andwhether or not it will be subject otthe client lib/ server hll | 23:08 |
| morganfainberg | Yes. | 23:08 |
| mordred | hell | 23:08 |
| mordred | but you know more than me | 23:08 |
| morganfainberg | The client server hell? | 23:08 |
| morganfainberg | I need a little more context. | 23:08 |
| dtroyer | pulling keystoneauth out of ksc doesn't change anything in the version dependency dept if both clients and servers use it | 23:08 |
| dtroyer | re master vs stable branches | 23:08 |
| mordred | morganfainberg: installation of master clients on stable servers breaking things | 23:08 |
| morganfainberg | Unfortunately, yes it will be used by both. But the hope is we can get the depends very very very very very narrow. | 23:09 |
| dtroyer | sigh | 23:09 |
| morganfainberg | But where it will be used by both is via clients | 23:09 |
| morganfainberg | Not direct by the server. | 23:09 |
| dtroyer | so how do we do stable/kilo with liberty-era clients? | 23:10 |
| *** chlong has joined #openstack-sdks | 23:10 | |
| mordred | ok. we need to sit down and figure this out at the summit | 23:10 |
| mordred | because I now how two conflicting desires | 23:10 |
| mordred | a) use openstacksdk | 23:10 |
| morganfainberg | Maybe we need to be smarter about this | 23:10 |
| mordred | b) use keystoneauth | 23:10 |
| dtroyer | seriously yes… this will put me over the edge and I'll just work on Go clients, or any other language without this damn installation nightmare | 23:10 |
| mordred | and that makes me sad panda | 23:10 |
| dtroyer | or we vendor it in multiple places and keep the vendored code minimal | 23:11 |
| morganfainberg | We might be able to dodge this with a minor split for a server version namespace. | 23:11 |
| dtroyer | or we put the version in the package name | 23:11 |
| mordred | I do _not_ want to use a second implementation of keystone auth client -that's just a silly idea - btu I also don't want dtroyer to quit | 23:11 |
| morganfainberg | I'm open to doing some smart namespace things to make this less icky. | 23:11 |
| morganfainberg | So we dodge the client server hell. | 23:12 |
| dtroyer | import keystoen.auth_1_2_7 is one answer… | 23:12 |
| morganfainberg | This is a new thing. Let's treat it like a new thing. | 23:12 |
| mordred | EW GROSS | 23:12 |
| dtroyer | so we can install them all ;) | 23:12 |
| mordred | well, I mean - I don't understand why liberty keystoneauth should break essex nova | 23:12 |
| morganfainberg | And not limit ourselves to "we sucked at this in the past" | 23:12 |
| mordred | and I think that if that is the case | 23:13 |
| mordred | then something is broken | 23:13 |
| morganfainberg | mordred: right. So how can we dodge that? And I don't want to vendor it. | 23:13 |
| dtroyer | we have to be able to install multiple versions of a thing. there are few ways to do it: venv/chroot/etc, versions in mnames, or vendor | 23:13 |
| mordred | keystone auth, of all the things, MUST handle ALL of the versions and have a 100% unchanging 100% compat python interface | 23:13 |
| morganfainberg | mordred: that is the plan with the 1.x series. | 23:13 |
| dtroyer | I would love to be missing something from that list | 23:13 |
| mordred | dtroyer: why do we have to install multiple versions of things? | 23:13 |
| *** dstufft has quit IRC | 23:14 | |
| morganfainberg | If we move to a 2.x I plan on making them install side-by-side | 23:14 |
| dtroyer | to support stable/* branches and still be able to install a client on the server itself | 23:14 |
| dtroyer | or devstack | 23:14 |
| mordred | if keystoneauth releases a 1.0 and never releases a new version, then we should not have the upgrade churn issue we had with python-*clent where people kept chasing pretty | 23:14 |
| mordred | yah | 23:14 |
| morganfainberg | 1.x of ksa can grow fixes but not new crazy deps. | 23:15 |
| mordred | zebra keystone auth should be able to be installed on a server running stable/liberty nova and nothing should break or else we shoudl all go home | 23:15 |
| morganfainberg | It is meant to be *stable* | 23:15 |
| morganfainberg | like *really* stable. | 23:15 |
| mordred | it has to be | 23:15 |
| mordred | or else it's useless | 23:15 |
| morganfainberg | Yes. | 23:15 |
| dtroyer | so the auth plugin objects need to be finished | 23:15 |
| morganfainberg | I'm planning on ripping all the g-r madness out of it too. | 23:15 |
| morganfainberg | dtroyer: plugins are hopefully always going to be separate packages besides the very basic ones. And again. Very limited requires. | 23:16 |
| dtroyer | I hope so…it feels like the base interface is still changing though. maybe I'm not reading jamie's reviews closely enough | 23:17 |
| morganfainberg | Seriously, I want ksa to be like libc, if you install v1 or v2 they can both play nice. | 23:17 |
| dtroyer | it will have to | 23:17 |
| dtroyer | thanks | 23:17 |
| morganfainberg | dtroyer: I expect it to change until 1.x | 23:17 |
| morganfainberg | The 0.1.0 release is so we can get feedback and start fixing things. | 23:17 |
| *** dstufft has joined #openstack-sdks | 23:17 | |
| *** dstufft has left #openstack-sdks | 23:17 | |
| mordred | dtroyer: ++ | 23:17 |
| *** shakamunyi has joined #openstack-sdks | 23:18 | |
| *** barra204 has joined #openstack-sdks | 23:18 | |
| morganfainberg | Once we cut 1.x new interfaces that break any compat is 2.x target | 23:18 |
| morganfainberg | Including anything that could be dependency changes. | 23:18 |
| morganfainberg | That is why we will be ripping Oslo things out of it. | 23:19 |
| morganfainberg | Pbr is my only real question mark. | 23:19 |
| dtroyer | I'll wait to look at 0.1.0 before jumping about over the dependencies ;) | 23:19 |
| dtroyer | yay! | 23:19 |
| dtroyer | (no oslo) | 23:19 |
| morganfainberg | Yeah 0.1.0 has extra deps. But that is because we wanted the cleanup in Gerrit. | 23:19 |
| dtroyer | not to dis Oslo, but they're meant for servers | 23:20 |
| morganfainberg | Not before import to Gerrit. | 23:20 |
| morganfainberg | dtroyer: exactly. | 23:20 |
| dtroyer | ok, cool. morganfainberg I need to buy you a beer next week for calming me down, thanks | 23:20 |
| morganfainberg | In fact, let me make an infra change so I can have a separate ksa core team. | 23:21 |
| dtroyer | I knew ksa would be small, light, etc, just not that it was planned for the SDK too | 23:21 |
| morganfainberg | It'll inherit keystone but maybe someone like you is the right person to add. | 23:21 |
| morganfainberg | Get SDK/client focus people also weighing in. | 23:22 |
| morganfainberg | dtroyer: it is planned to supplant the awful mess of needing ksc to do auth. | 23:22 |
| dtroyer | I'd be happy to do that | 23:22 |
| morganfainberg | Log term, I want bindings for non-putting langs for it too. | 23:22 |
| morganfainberg | Long* | 23:22 |
| morganfainberg | python* | 23:23 |
| morganfainberg | Damn autocorrect. | 23:23 |
| morganfainberg | As soon as I am home I'll make the infra proposal change. | 23:23 |
| dtroyer | FWIW, I have already started re-working the existing go lib to use the session-style so things ought to translate easier | 23:23 |
| morganfainberg | I really want ksa to be an example of the right way to isolate specific functionality. | 23:23 |
| morganfainberg | dtroyer: yeah step away from the cliff. We're here to make your life better :) | 23:25 |
| morganfainberg | jamielennox|away: ^^ btw - see what I just committed you to? :P | 23:25 |
| morganfainberg | *shiftyeyes* | 23:26 |
| dtroyer | I know, and thanks again. The madness must be stopped one way or another | 23:26 |
| morganfainberg | So let's take a look at pbr. We might need to ditch it for ksa | 23:27 |
| morganfainberg | Talk at the summit about that? Ksa will need to not be proposal bot requirements updated. | 23:27 |
| dtroyer | I have no opinion there unless/until the rest of the client stack doesn't need it too | 23:27 |
| morganfainberg | Well the issue is pbr could cause compat issues in edge cases here. | 23:28 |
| morganfainberg | I'll need to corner mordred and lifeless on that. | 23:28 |
| morganfainberg | If we need non-pbr setup for ksa, we can do that. | 23:28 |
| morganfainberg | But i want to make sure we aren't cornering ourselves in either case. | 23:29 |
| mordred | we dn't need it to be a requirements requirement | 23:31 |
| mordred | build requirement is fine | 23:31 |
| mordred | and shouldnt' be a problem | 23:31 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for database proxy gets https://review.openstack.org/182788 | 23:33 |
| morganfainberg | mordred: ok cool. | 23:33 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 23:33 | |
| morganfainberg | I think requests and six are going to be the hard-to-avoid requirements. | 23:33 |
| morganfainberg | And argparse *glare Python 2.6* | 23:34 |
| mordred | morganfainberg: why do you need argparse in keystoneauuth? | 23:34 |
| mordred | it's a library, not a commnad line program | 23:34 |
| morganfainberg | mordred: there is some option handling. Not sure. Hope to drop that too. | 23:34 |
| mordred | drop it | 23:34 |
| morganfainberg | All the option handling that is. | 23:34 |
| mordred | it's not appropriate | 23:34 |
| mordred | it should have a library interface that's clean | 23:34 |
| morganfainberg | But it's part of the cleanup. Ideally, six and requests. | 23:35 |
| morganfainberg | And stevedore. | 23:35 |
| morganfainberg | For plugins. | 23:35 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for get calls in image proxies https://review.openstack.org/182804 | 23:35 |
| morganfainberg | That would make me very happy. | 23:35 |
| mordred | morganfainberg: thereyago: https://review.openstack.org/183352 no pbr :) | 23:36 |
| morganfainberg | Ahaha. Sounds good. | 23:36 |
| *** pm90_ has quit IRC | 23:38 | |
| mordred | morganfainberg: otoh - argparse isn't ACTUALLY a problem since it's built in to 2.7 and beyond | 23:38 |
| mordred | and even the 2.6 lib has a very stable interface | 23:38 |
| morganfainberg | Aha right. | 23:39 |
| morganfainberg | But still silly to need it. | 23:39 |
| openstackgerrit | Merged stackforge/python-openstacksdk: Changes for compute proxy list calls https://review.openstack.org/183206 | 23:39 |
| dtroyer | IIRC argparse was needed to do all of the oslo.config stuff that I think could be layered on top if it is still required. I agree it doesn't belong in ksa | 23:43 |
| *** pm90_ has joined #openstack-sdks | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!