| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-sdks - 2017-10-13 | |||
| 20:49:18 | dtroyer | oh I need a laugh | |
| 20:49:29 | mordred | dtroyer: in pagination, next links include the API version - so you get "next": "/v2/images?marker=bb15732f-e016-4544-a7e9-c73db58ce3e5" | |
| 20:49:58 | dtroyer | wheeee!!!!!!!!! | |
| 20:50:02 | mordred | dtroyer: which would be fine - except that the base catalog url is usually 'https://images.example.com/v2' | |
| 20:50:14 | dtroyer | to be fair, pagination is a general mess almost everywhere | |
| 20:50:27 | mordred | so if youhave an Adapter on https://images.example.com/v2 and you follow the next link ... you get https://images.example.com/v2/v2/images?marker=bb15732f-e016-4544-a7e9-c73db58ce3e5 | |
| 20:50:30 | mordred | \o/ | |
| 20:50:35 | mordred | dtroyer: indeed | |
| 20:51:00 | dtroyer | but that one is special… | |
| 20:51:24 | mordred | yah. it's even internally inconsistent - that's what I really like about it | |
| 20:52:48 | mordred | I believe part of the issue is that glanceclient strips the version suffix from the catalog and stores the url internally as https://images.example.com - then makes all of its REST calls as 'GET /v2/images' ... so for glanceclient the dissonance doesn't surface | |
| 20:53:06 | mordred | dtroyer: anywho - I thought you'd enjoy that one :) | |
| 20:54:31 | dtroyer | glanceclient is, uh, well you know… | |
| 20:56:27 | cdent | mordred: is that a problem with the next link or the adapter? | |
| 20:56:42 | cdent | presumably the adapter should see an absolute link as an absolute link? | |
| 20:57:02 | cdent | I have a similar problem when dealing prefixes in gabbi, it’s a pain in the ass | |
| 20:57:05 | mordred | cdent: well... it's not always that simple | |
| 20:57:06 | cdent | there’s no great solution | |
| 20:57:12 | mordred | cdent: bcause of suburl services | |
| 20:57:41 | cdent | right, I know, that’s what I’m saying: adapters (not just ksa) need to have more brains | |
| 20:57:48 | cdent | which is a pain | |
| 20:57:50 | mordred | yah | |
| 20:58:09 | mordred | I'd say the bug is in the next link - because it should EITHER be 'images?marker=bb15732f-e016-4544-a7e9-c73db58ce3e5' | |
| 20:58:19 | mordred | in which case the correct behavior is easy to figure out | |
| 20:58:32 | mordred | or it shold be 'https://images.example.com/v2/images?marker=bb15732f-e016-4544-a7e9-c73db58ce3e5' | |
| 20:58:36 | mordred | in which case same thing | |
| 20:58:46 | cdent | would you accept ///images? | |
| 20:58:54 | cdent | or is it // (can’t remember) | |
| 20:59:25 | mordred | yah - ksa adapter would do the right thing with that ... but i'm not sure it's correct for glance to return that - because glance might or might not be deployed at a url root | |
| 20:59:45 | cdent | sorry, that wasn’t I mean | |
| 20:59:46 | cdent | t | |
| 21:00:22 | cdent | /v2/images[…] or ///v2/images[…] (whichever is actually the magic fully qualify me as in the current context) | |
| 21:00:51 | mordred | yah- so ... we currently treat /v2/images as 'fully qualify me in the current context' ... | |
| 21:01:13 | cdent | I think the relative is probably better in the long run, but *shrug* | |
| 21:01:15 | mordred | it's just that the current context in this case is https://images.example.com/v2 - which is why it appends like that | |
| 21:03:44 | cdent | sorry, I’m talking a browser thing where // (or /// can’t remember) means off root of whatever the scheme and host is of “this page” | |
| 21:03:49 | mordred | so - technically I think we're doing that join wrong | |
| 21:03:54 | cdent | yes | |
| 21:04:12 | mordred | >>> urllib.parse.urljoin('https://images.example.com/images/v2', '/v2/images') | |
| 21:04:14 | mordred | 'https://images.example.com/v2/images' | |
| 21:04:16 | mordred | >>> urllib.parse.urljoin('https://images.example.com/images/v2', '//v2/images') | |
| 21:04:18 | mordred | 'https://v2/images' | |
| 21:04:36 | cdent | yeah, I suspect urllib doesn’t speak browser magic | |
| 21:05:11 | mordred | yah - mainly - next links returning rooted/absolute urls is a thing we shold not do | |
| 21:05:19 | cdent | yah | |
| 21:05:42 | mordred | since that makes suburl deployments and version discovery roots both very ... confusing ... to deal with | |
| 21:06:04 | mordred | whereas if it was just 'next': 'images?marker=bb15732f-e016-4544-a7e9-c73db58ce3e5' | |
| 21:06:21 | mordred | then the joins will all work properly no matter what the setup is - and it's not unclear what it's relative to | |
| 21:06:30 | cdent | yah | |
| 21:07:34 | cdent | such typing. wow. | |
| 21:08:41 | cdent | I think I should exit this establishment in favor of some relaxation. Good night all. | |
| 22:10:52 | openstackgerrit | Dean Troyer proposed openstack/cliff master: Add a --errexit option to exit the comamnd loop on command errors https://review.openstack.org/511971 | |
| 22:19:53 | openstackgerrit | Merged openstack/cliff master: doc: Cleanup of demoapp doc https://review.openstack.org/510802 | |
| #openstack-sdks - 2017-10-15 | |||
| 20:43:22 | openstackgerrit | Daniel Speichert proposed openstack/python-openstackclient master: Allow creating security rules without protocol https://review.openstack.org/512113 | |
| 23:49:01 | openstackgerrit | Ian Wienand proposed openstack/python-openstackclient master: Switch to $USER in post_test_hooks https://review.openstack.org/512125 | |
| 23:50:31 | openstackgerrit | Ian Wienand proposed openstack/python-openstackclient master: Switch to $USER in post_test_hooks https://review.openstack.org/512125 | |
| 23:53:03 | openstackgerrit | Ian Wienand proposed openstack/python-openstackclient master: Move base functional test job in-repo https://review.openstack.org/509015 | |
| #openstack-sdks - 2017-10-16 | |||
| 04:10:30 | ldl | hi, friends, may I ask a question? Can I get the server's snapshots from the server? | |
| 07:39:21 | ldl | Hi, friends, may I ask a question? | |
| 07:40:02 | ldl | recover from the snapshot image, if is equals to the rebuild VM? | |
| #openstack-sdks - 2017-10-17 | |||
| 01:26:15 | openstackgerrit | Merged openstack/python-openstackclient master: Switch to $USER in post_test_hooks https://review.openstack.org/512125 | |
| 07:20:49 | openstackgerrit | Dinesh Bhor proposed openstack/keystoneauth master: Add mask_password to sanitize sensitive data https://review.openstack.org/512522 | |
| 14:10:52 | openstackgerrit | Merged openstack/python-openstackclient master: Move base functional test job in-repo https://review.openstack.org/509015 | |
| 15:54:56 | openstackgerrit | Daniel Speichert proposed openstack/python-openstackclient master: Allow creating security rules without protocol https://review.openstack.org/512113 | |
| 15:56:48 | openstackgerrit | Daniel Speichert proposed openstack/python-openstackclient master: Allow creating security rules without protocol https://review.openstack.org/512113 | |
| 16:03:38 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Move more jobs in-repo https://review.openstack.org/512680 | |
| 16:26:55 | openstackgerrit | Vladyslav Drok proposed openstack/osc-lib master: Allow to use the none auth plugin https://review.openstack.org/512699 | |
| 17:38:01 | openstackgerrit | Daniel Speichert proposed openstack/python-openstackclient master: Allow creating security rules without protocol https://review.openstack.org/512113 | |
| 18:20:08 | openstackgerrit | Vladyslav Drok proposed openstack/osc-lib master: Allow to use the none auth plugin https://review.openstack.org/512699 | |
| 18:32:33 | openstackgerrit | Vladyslav Drok proposed openstack/osc-lib master: Allow to use the none auth plugin https://review.openstack.org/512699 | |
| 19:34:37 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Move more jobs in-repo https://review.openstack.org/512680 | |
| 22:25:31 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Move more jobs in-repo https://review.openstack.org/512680 | |
| #openstack-sdks - 2017-10-18 | |||
| 00:08:47 | openstackgerrit | Dean Troyer proposed openstack/cliff master: Add an errexit attribute to InteractiveApp to exit on command errors https://review.openstack.org/511971 | |
| 00:10:06 | openstackgerrit | Dean Troyer proposed openstack/osc-lib master: Add --errexit option to pass to interactive mode https://review.openstack.org/512903 | |
| 02:56:05 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Move more jobs in-repo https://review.openstack.org/512922 | |
| 02:58:42 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Move more jobs in-repo https://review.openstack.org/512680 | |
| 03:04:05 | openstackgerrit | Dean Troyer proposed openstack/python-openstackclient master: Move more jobs in-repo https://review.openstack.org/512680 | |
| 12:25:19 | openstackgerrit | Monty Taylor proposed openstack/os-client-config master: Update make_rest_client to work with version discovery https://review.openstack.org/513013 | |
| 12:25:41 | mordred | Shrews, dtroyer: ^^ discovered that issue today in a workshop I was giving | |
| 12:26:53 | mordred | dtroyer: also, if you get a sec, https://review.openstack.org/#/c/506194/ and https://review.openstack.org/#/c/505388/ could use some love | |
| 13:13:44 | Shrews | mordred: minor issue with the reno note, otherwise lgtm | |
| 13:35:54 | openstackgerrit | Dean Troyer proposed openstack/openstackclient master: Add OSC plugin job https://review.openstack.org/509012 | |
| 13:49:27 | openstackgerrit | Monty Taylor proposed openstack/os-client-config master: Update make_rest_client to work with version discovery https://review.openstack.org/513013 | |
| 13:50:54 | mordred | dtroyer: zomg. when did the python-openstackclient repo change name??? | |
| 13:53:15 | dtroyer | it idin't | |
| 13:53:27 | dtroyer | that's a meta-project that also installs all of the known plugins | |
| 13:53:50 | dtroyer | it hasn't been released yet, and is where the plugin tests should ultimately live | |
| 13:54:41 | mordred | ah - gotcha | |
| 13:54:46 | mordred | dtroyer: I don't see check-osc-plugins.sh in the repo | |
| 13:54:58 | dtroyer | damn | |
| 13:55:23 | dtroyer | I've been fighting the plugin job in osc, thought moving here would be quicker… there's always something :) | |
| 13:55:27 | mordred | :) | |
| 13:56:12 | dtroyer | ok, so since you're here, what is the right way to find that script? ie, what var do I prefix the path 'src/git.openstack.org/…' with? | |
| 13:56:29 | mordred | WELL ... I was just writing you a patch | |
| 13:57:11 | mordred | dtroyer: the legacy jobs are set up to chdir to ~/workspace - but that's not really a zuulv3 thing -it's just a thing we put into the base legacy job for transition purposes | |
| 13:57:24 | mordred | dtroyer: (I'm answering you extra verbose) | |
| 13:58:06 | mordred | dtroyer: but it doesn't seem you actually need the legacy bits - so we can strip them, at which point your current working dir should be /home/zuul ... meaning referencing src/git.openstack.org/openstack/openstackclient should work fine | |
| 13:58:23 | mordred | however - you can ALSO reference it via ~/src/git.openstack.org/openstack/openstackclient if you prefer | |
| 13:58:48 | dtroyer | I was trying to do this in steps so understand the transition from legacy to native better | |
| 14:01:28 | mordred | nod. would you find it helpful or counter-productive for me to push up a followup patch to your patch that de-legacies it? (I wanna be helpful in whichever way is more helpful) | |