| === harlowja is now known as harlowja_away | ||
| === alexpilotti_ is now known as alexpilotti | ||
| smoser | cmiN, and claudiupopa happy to have you all here. | 15:12 |
|---|---|---|
| === harlowja_away is now known as harlowja | ||
| === harlowja is now known as harlowja_away | ||
| === harlowja_away is now known as harlowja | ||
| smoser | harlowja: around ? | 20:10 |
| harlowja | yo | 20:10 |
| harlowja | sup | 20:10 |
| smoser | why would/should i like to use testr | 20:10 |
| smoser | versus something else. | 20:10 |
| harlowja | i personally would say don't, lol | 20:10 |
| harlowja | i personally don't like it :-P | 20:10 |
| harlowja | it does running tests in parallel better | 20:11 |
| smoser | so lets say that you were suggesting something... what would you suggest ? | 20:11 |
| harlowja | but its output imho suck | 20:11 |
| smoser | just nose ? | 20:11 |
| harlowja | ya, i still think its adeqaute; although pytest is supposedly nice also (although i haven't used it) | 20:11 |
| harlowja | http://pytest.org/latest/ | 20:12 |
| harlowja | i just don't like testr and its creation of weirdly formatted files that are hard to undersetand | 20:12 |
| harlowja | crap like http://logs.openstack.org/10/148810/1/check/gate-taskflow-python27/8b0c041/subunit_log.txt.gz | 20:13 |
| harlowja | but maybe thats really more subunits fault; but testr seems to use it so i'll blame testr, lol | 20:13 |
| harlowja | so imho testr brings in alot of other projects (which sorta sux) | 20:14 |
| harlowja | testttools, subunit, testr... | 20:14 |
| harlowja | i think nose though is sorta dying/not updated anymore | 20:16 |
| harlowja | nm, thats a lie, ha | 20:16 |
| smoser | harlowja: https://github.com/cloud-init/cloud-init | 21:50 |
| smoser | fyi. initial stub directory tree. | 21:50 |
| smoser | now to make something actually work . :) | 21:50 |
| harlowja | woot | 22:01 |
| harlowja | looks good to me, ship it | 22:02 |
| larsks | Is that replacing the bzr repository at lp:cloud-init? | 22:02 |
| harlowja | i thinks so | 22:02 |
| larsks | Spiffy. | 22:03 |
| harlowja | now i guess is the question of how do we want this to operate? will this be a daemon that is connected into systemd events (and modules get triggered on events?) | 22:05 |
| harlowja | or will it retain the CLI activation? | 22:05 |
| harlowja | (or both or other...) | 22:05 |
| larsks | You mean, for a next-generation version of cloud-init? | 22:06 |
| harlowja | ya | 22:06 |
| harlowja | *since afaik thats what the above will become | 22:06 |
| larsks | Huh. I would argue for not linking it too tightly to systemd, so that it is easy to support it in other environment (other linux distros/freebsd/solaris/etc). | 22:07 |
| larsks | Unless the goal is really to make it only-distributions-running-systemd sort of thing. | 22:07 |
| harlowja | i hope not, but it would seem like for linux responding to events (network plugged, reset the network information, new-metadata-arrived or something...) would be an approach | 22:08 |
| harlowja | guess it depends on where we want to go here :-P | 22:08 |
| larsks | Sure. So we can design cloud-init to make that easy through systemd units that run various command lines. | 22:09 |
| larsks | Things I want: | 22:09 |
| larsks | - cache metadata from the cloud provider, and then provide a query interface | 22:09 |
| larsks | - stop masking exceptions, which makes it really hard to debug things | 22:09 |
| harlowja | sure, seems reasonable | 22:10 |
| harlowja | larsks add stuff to https://etherpad.openstack.org/p/cloud-init-next if u want | 22:12 |
| larsks | Ooo, shiny. | 22:12 |
| harlowja | *if u haven't already | 22:12 |
| larsks | Are the stackforge suggestions yours? | 22:14 |
| harlowja | i don't think so, ha | 22:14 |
| harlowja | but might be :-P | 22:14 |
| larsks | Because that seems like a really good idea, too. | 22:14 |
| harlowja | nah, spelling is to good to be mine | 22:15 |
| harlowja | so def not mine, lol | 22:15 |
| harlowja | btw, metadata is already cached locally | 22:15 |
| harlowja | http://cloudinit.readthedocs.org/en/latest/topics/dir_layout.html | 22:16 |
| harlowja | the metadata should be pickled into 'datasource' there (which is the datasource object) | 22:16 |
| harlowja | maybe not as visible as it could be though | 22:16 |
| larsks | Yeah. | 22:16 |
| larsks | And picke is awful python-specific. | 22:16 |
| larsks | Providing a cli for querying it might help, though. | 22:16 |
| harlowja | sure | 22:17 |
| harlowja | reminds me of https://code.launchpad.net/~harlowja/cloud-init/query-back-duo :-P | 22:17 |
| harlowja | from back in the day, ha | 22:17 |
| harlowja | or https://code.launchpad.net/~harlowja/cloud-init/query-tool-is-back ha | 22:18 |
| harlowja | didn't know there was 2 of those, lol | 22:18 |
| larsks | Nice. | 22:18 |
| larsks | The cirros cloud-init stuff does provide a query tool, which is nifty. | 22:18 |
| harlowja | cool | 22:18 |
| harlowja | cubswin | 22:18 |
| harlowja | lol | 22:18 |
| harlowja | blame smoser for that password, lol | 22:19 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!