[08:40] <achilleasa> jam: I have pushed some commits to the charm.v6 PR to make the tests easier to read. If it looks good I will push a cleaned up commit log and merge
[08:41] <stickupkid> jam, you got 5 minutes to discuss the following https://github.com/juju/python-libjuju/pull/321#discussion_r305900673
[08:41] <jam> stickupkid: sure, I need about 10min then we can chat
[09:01] <jam> achilleasa: lgtm
[09:01] <jam> stickupkid: heading to HO
[11:43] <stickupkid> jam, here is the pinned version of the facades https://github.com/juju/python-libjuju/pull/321/commits/3ecc14fcc65bebfcebe402892b7074b14c2fa4f7
[11:44] <stickupkid> jam, the client now very roughly (v1) states which version it knows how to talk about - if you want the latest and greatest, you have to bump this and understand the changes
[12:10] <achilleasa> Can I get a quick CR on https://github.com/juju/charm/pull/285? It's just a spelling typo fix
[12:21] <manadart> hml: All caught up with rebasing/merging. Latest one is for review here: https://github.com/juju/juju/pull/10455
[12:22] <hml> manadart:  good news. looking
[12:26] <hml> manadart: did my pr get closed when the merge completed on your pr?
[12:28] <manadart> hml: Yeah: https://github.com/manadart/juju/pull/1
[12:29] <hml> manadart: darn it, it won’t let me change the  backing.  oh well
[12:30] <hml> manadart:  i can review for you.  it’d be easier if someone else could qa though
[12:40] <achilleasa> hmmm bringing my charmrepo.v5 changes into juju breaks stuff as some of the code needs macaroon-bakery.v2 and other parts need macaroon-bakery.v2-unstable... does anyone have a bit of time to pair with me to help?
[12:51] <jam> achilleasa: presence lgtm
[12:52] <jam> stickupkid: minor tweak in that we need a way to specify that we could support >1 version. so the version pointer needs to be a list, and we should pick the highest version that is in both lists.
[12:52] <jam> stickupkid: otherwise, I like it a lot
[12:53] <stickupkid> jam, right yeah, ok makes sense
[12:55] <jam> stickupkid: I think it is "max(set(known).intersect(set(discovered)))" with caveats for things like "no set overlap"
[12:55] <jam> unlikely to happen in practice so we can be a bit jankier there as long as seeing a facade we don't know about doesn't kill us from using all the ones we *do* know about
[12:56] <jam> stickupkid: speaking of, should we set "facades[-1]" for ones we don't know, or should we just pretend they don't exist ?
[12:56] <stickupkid> jam, nice, ok
[12:56] <stickupkid> jam, so i think they shouldn't exist, but people use the clients directly so i'm unsure
[12:57] <jam> stickupkid: k, so to the Shim, I would say they don't exist. to the list of what is available for someone using a client directly I'm less concerned. It would be good to have something like "available facades"
[12:57] <stickupkid> agreed
[12:58] <jam> all the versions the libjuju could give you intersected with the versions the server supports
[12:58] <stickupkid> yeah
[13:10] <jam> stickupkid: 10453 reviewed
[13:13] <hml> stickupkid: there were some FAILs in the test with the weird make end.  trying again
[13:15] <stickupkid> jam, replied https://github.com/juju/juju/pull/10453
[13:35] <stickupkid> jam, https://github.com/juju/python-libjuju/pull/321/commits/e4d7279f91e1792a3b1afd5bab1b7fe3735c4a96 pylibjuju now picks the best between what it knows and what the api server knows \o/
[14:04] <hml> manadart:  slice of az is now rebased and ready for review:  https://github.com/juju/juju/pull/10456
[14:05] <manadart> hml: OK.
[14:16] <manadart> hml: Trade you for forward-merge: https://github.com/juju/juju/pull/10458
[14:18] <hml> manadart:  i have one more pr for 2.6 if you don’t mind waiting an hour.  had to fix a merge conflict,  pushing up now to merge the pr.  :-D
[14:18] <hml> manadart: and yes i’ll review the merge into develop
[14:31] <stickupkid> ah damn it, not this again go install github.com/juju/juju/acceptancetests/repository/trusty/fill-logs/actions: build output "/home/ubuntu/go/bin/actions" already exists and is not an object file
[14:47] <stickupkid> hml, you're right https://github.com/juju/juju/pull/10424
[14:49] <hml> stickupkid: the bug says it’s reproducable in 2.6 which makes me wonder about the clouds.yaml and credentials.yaml.
[14:49] <hml> stickupkid: i never added credentials for local anywhere
[14:49] <stickupkid> hml, yeah... i did fix this when i re-wrote the LXD credentials stuff a long time ago, so was surprised it broke
[14:51] <stickupkid> hml, i think something broke 2.5 and then anastasia fixed it with her credentials work for 2.6
[14:51] <stickupkid> hml, maybe not worth fixing tbh
[14:54] <hml> manadart:  sequence for spaceid is on a per model basis yes?
[14:54] <manadart> hml: Yes, same collection/mechanism as machine IDs etc.
[14:55] <hml> manadart: does the sequence for a model migrate too?
[14:57] <manadart> hml: Yes. The sequences collection is imported early, so everything works - later migration logic can add entities with the next available IDs and such.
[15:03] <hml> manadart:  if i migrate a model from develop to pr base controller, the spaces ids go from 1-4 to 5-8.  if i migrate again from a pr based controller to other pr based controller, the ids stay 5-8.
[15:09] <manadart> hml: Yes. Because develop got the IDs before juju/description was updated, so they were not transported in the serialised model.
[15:09] <manadart> So new ones get created at the other end.
[15:10] <manadart> Migrations between stable releases should be OK though.
[15:10] <hml> manadart:  right… between stable releases they stay the same.
[15:10]  * manadart nods.
[15:11] <manadart> hml: Approved the AZ patch too.
[15:11] <hml> manadart: approved the merge
[15:11] <hml> ty
[15:12] <manadart> hml: Ta.
[15:23] <stickupkid> hml, fixed 2.5 issue :) https://github.com/juju/juju/pull/10459
[15:24] <hml> stickupkid: ack - will look at it this afternoon if that’s okay
[15:24] <stickupkid> hml, fine by me, if it works can you $$merge$$ it for me
[15:24] <hml> stickupkid: :-)
[15:25] <stickupkid> hml, ta
[17:24] <achilleasa> can I get a CR on https://github.com/juju/juju/pull/10460? This touches the deploy code so please take your time and try to break it ;-)