[01:33] <wallyworld> babbageclunk: were you looking at the sig file stuff for beta2 release?
[01:33] <babbageclunk> yup, just finished it off - I'll put the PR up while I'm doing a manual test.
[01:33] <wallyworld> yay, ty
[01:59] <axw> wallyworld: small fix for offer removal: https://github.com/juju/juju/pull/7976
[01:59] <wallyworld> ty, looking
[02:00] <babbageclunk> wallyworld: https://github.com/juju/juju/pull/7977 as well
[02:08] <wallyworld> babbageclunk: ty, looking after my 1:1 with andrew
[02:08] <babbageclunk> cool cool
[02:24]  * babbageclunk goes for a run
[02:26] <wallyworld> babbageclunk: just a small test change
[03:28] <babbageclunk> wallyworld: thanks, fixed - merging now.
[04:05] <wallyworld> babbageclunk: no worries
[04:30] <anastasiamac> thumper: wallyworld: axw: modelstatus fix - https://github.com/juju/juju/pull/7979... since it's kind of not-small and highly visible/usable, I think I'd need at least 2 reviews...
[04:49] <axw> anastasiamac: looking after lunch
[04:50] <anastasiamac> axw: tyvm
[04:56] <wallyworld> anastasiamac: +631/-179 is but a babe in the woods compared to some lately :-)
[04:57] <anastasiamac> wallyworld: m glad to hear that my PR does not scare u... m not sure how i feel about the fact that big PRs r becoming more frequent
[04:57] <wallyworld> :-)
[05:04] <jam> anastasiamac: revewed
[05:04] <anastasiamac> jam: \o/ thnx
[05:04] <jam> morning
[05:05] <jam> doesn't seem like a PR that needs 2 reviews
[05:06] <anastasiamac> jam: k, m happy(ier) to have just one review (ur definition of weekend is very different to mine)
[05:08] <jam> anastasiamac: what, I woke up, like an hour later, went to the grocery store, will be going out for lunch
[05:08] <jam> how is that not a weekend?
[05:08] <anastasiamac> jam: bestversion check on cmd is needed for cases where u r using new client against an older controller (if it ever happens) as we'd come to result.Error and will panic/err since Error would not exist in returned results...
[05:09] <jam> anastasiamac: no, because your api/* code forces a result.Error => result[i].Error, doesn't it?
[05:09] <jam> so you've already made sure that every object has a .Error set
[05:16] <wallyworld> jam: i'd like to discuss your +1 to the pr - i don't agree with the approach
[05:17] <wallyworld> i can't see a reason to not have the api layer present the same interface for either a v3 or v4 backend
[05:17] <wallyworld> we control the call sites for the api layer so we can make everything hang othether as it should
[05:19] <wallyworld> anastasiamac: jam: axw: happy to discuss in a hangout to ensure we're all on the same page, maybe easier than irc
[05:20] <anastasiamac> wallyworld: +1, gimme 10mins?
[05:20] <wallyworld> sure
[05:21] <jam> wallyworld: I thought she did have the api/ layer push the global error into the each object Error
[05:21] <jam> did I misread that?
[05:21] <wallyworld> maybe i did, left me check
[05:22] <wallyworld> jam: see line 84 in processModelStatusResultsV3
[05:22] <jam> line 61 of processModelStatusResults
[05:22] <jam> the part where we were talking about Trace
[05:23] <wallyworld> that's for the v4 facade
[05:23] <wallyworld> not the old v3
[05:25] <jam> wallyworld: ah k. in that case, update v3 to set the error, and I think we're happy
[05:25] <wallyworld> modulo that we don't need a separate processv3 method perhaps
[05:26] <wallyworld> since v3 will just return an overall error anyway
[05:26] <wallyworld> and if we do get results we just process them with the one method
[05:28] <wallyworld> and we need to check but i thought jimm calls this api also so we'll need to update that code as well, or organise for it to be updated
[05:28] <jam> so I think we're roughly aligned, in that api/* should provide a consistent view of errors, and then all the client code doesn't need BestVersion.
[05:28] <jam> wallyworld: as a Controller facade, it should have some sort of implementation for it
[05:28] <wallyworld> right, so we'll break them if they compile against tip
[05:28] <jam> wallyworld: though if they did it correctly, we just treat JIMM as an older controller
[05:28] <anastasiamac> wallyworld: jam: axw: want to take it to team standup from this morning?
[05:28] <wallyworld> or in any case, we'll need to get them to update
[05:29] <wallyworld> sure, ok
[05:29] <jam> https://hangouts.google.com/hangouts/_/canonical.com/juju-standup?authuser=1 ?
[05:29] <anastasiamac> jam: yes
[05:29] <anastasiamac> wallyworld: ?
[16:55] <ryebot> What happened to the --upload-tools flag on the juju bootstrap command? What's the right way to replicate that functionality today?
[23:20] <dvavili_> Not sure if it's the right channel, but can Juju be used to deploy P3 instances on AWS that was announced a couple of days back?