[00:01] <babbageclunk> Fun test fix: https://github.com/juju/juju/pull/11014
[00:03] <lucidone> It doesn't look like the juju gui supports SAAS, and the UI crashes if you try to show relations on an application that is related to a SAAS application .. Is https://bugs.launchpad.net/juju-gui the place to report a bug?
[00:30] <babbageclunk> lucidone: yes, that would be the right place to start (although it might turn out to be a juju bug under the hood)
[00:32] <lucidone> babbageclunk: Cheers, this bug in particular is a js error "TypeError: Cannot read property 'serviceName' of undefined", so I imagine it's in the gui code. Supporting SAAS in juju gui itself feels more like a feature request
[00:33] <babbageclunk> lucidone: yeah, that makes sense
[01:05] <wallyworld> babbageclunk: state pr lgtm with a small fix to error processing
[01:05] <babbageclunk> wallyworld: cool, thanks
[01:17] <lucidone> Looked like there was more activity on github issues (and I was blocked by ubuntuone login not working on launchpad..) so I've posted to github https://github.com/juju/juju-gui/issues/4023
[01:53] <babbageclunk> lucidone: oops, sorry for the bum steer there
[01:54] <lucidone> babbageclunk: No worries! I did see a 2019 bug report there, so didn't appear to be totally defunct
[02:29] <timClicks> is it possible to serve resources locally, or is the charm store required?
[02:29] <lucidone> As in $ juju deploy /path/to/charm ?
[02:30] <timClicks> lucidone: I am developing a charm that needs a resource
[02:30] <timClicks> but I don't want to run "charm attach" yet to upload it to the charm store (it's not ready)
[02:31] <lucidone> Ah right
[02:31]  * lucidone reads the docs for resources
[02:32] <timClicks> it looks like I can use `juju attach-resource`
[02:58] <ec0> thumper: if you're around in about 30 minutes I should be free to catch up on that Cisco upgrade
[02:58] <ec0> disregard
[03:03] <babbageclunk> timClicks: I'm late to the party but I think you can use deploy with --resource to attach the resource at deploy-time as well (rather than doing it after the fact)
[03:04] <timClicks> babbageclunk: yeah you can :)
[03:04] <babbageclunk> phew
[07:58] <pal> Good Morning Juju!
[09:25] <nammn_de1> ptal, small one: https://github.com/juju/juju/pull/11011
[09:25] <nammn_de1> *ptal, please? :D
[09:26] <nammn_de1> stickupkid: morning, thats what i meant yesterday with forcing in init. Sadly we cannot use supercommand fields, thus no commonflags. https://github.com/juju/cmd/pull/71
[09:26] <nammn_de1> stickupkid: wanted to discuss that before adding more tests ect.
[09:31] <stickupkid> manadart, ping
[09:32] <manadart> stickupkid: Pong.
[09:32] <stickupkid> ho?
[09:33] <manadart> OMW. 2 secs.
[09:36] <nammn_de1> stickupkid: hows that comment? https://pastebin.canonical.com/p/xq2ShFXnvY/
[09:36] <stickupkid> nammn_de1, ok, that'll do
[09:40] <nammn_de1> stickupkid: great, updated pr
[09:50] <nammn_de1> stickupkid manadart rick_h: just realized as i was going through the branches test why they fail on 2.7. Do we want to backport this? https://github.com/juju/juju/pull/10796/
[09:51] <nammn_de1> nvm me, seems like they are..
[10:04] <nammn_de1> stickupkid: for the pr https://github.com/juju/juju/pull/11011 i updated the comments+naming as you said, additionally i added the fix to make sure that there is no race condition for the acceptancetest test-branches
[10:39] <manadart> nammn_de1: Yes, back-port it.
[10:44] <nammn_de1> manadart: apologies, i just realised this already has been the case. Lucky me
[10:45] <nammn_de1> while at it, manadart stickupkid: ptal https://github.com/juju/juju/pull/11011 ?
[11:30] <nammn_de1> stickupkid: if you have time later we can talk/look at the init change here https://github.com/juju/cmd/pull/71 . Ping me up if you up for it
[11:31] <nammn_de1> how do i enable all featureflags?
[11:31] <nammn_de1> is it exporting `export JUJU_DEV_FEATURE_FLAGS="" `?
[11:34] <nammn_de1> manadart: changed it, you will hate me, but i rebased as this change was more trivial, hope you don't mind:  https://github.com/juju/juju/pull/11011/files
[11:57] <stickupkid> manadart, CR on this one please :D https://github.com/juju/description/pull/69
[11:59] <stickupkid> ta
[14:07] <nammn_de1> stickupkid: ping me if you are around for my python questions and/or the init cmd thingy
[14:08] <stickupkid> nammn_de1, do it now?
[14:08] <nammn_de1> stickupkid sure, daily?
[14:08] <stickupkid> yeap
[14:26] <manadart> stickupkid: https://github.com/juju/juju/pull/11020. I am heading to HO.
[15:26] <nammn_de1> stickupkid: what was the output of doing it in init again? How do we want to proceed? https://github.com/juju/cmd/pull/71
[15:45] <nammn_de1> rick_h: im a bit struggling to be able to connect to the allwatchersapi through python, any pointers for me?  `WatchAllModels` returns an id of  an allwatcher which iam supposed to use with something
[15:45] <nammn_de1> just cant find something easiliy
[15:58] <rick_h> nammn_de1:  sec, otp for a bit can help look afterwards
[16:03] <nammn_de1> rick_h: thanks! stickupkid: is there a way to set a context for a facade connection?
[16:03] <nammn_de1> as it seems the watcher_id is taken from a context param, which i cannot give for the watcher.next() call
[16:04] <stickupkid> nammn_de1, have a look at what the go code, see how it uses it, that's generally what I'd do...
[16:16] <nammn_de1> cr someone? https://github.com/juju/juju/pull/11021
[16:59] <nammn_de1> stickupkid: they only api things  we can call through python have to be behind facades, right?
[16:59] <stickupkid> yes
[17:00] <nammn_de1> so this call https://github.com/juju/juju/blob/6ea20273f7c540d93b1075d77aafe39e70770b33/api/controller/controller.go#L186, which is returning the real watcher, is not directly possible and we might have a facade equivalent construct somewhere?
[17:02] <stickupkid> nammn_de1, not at all, each go client knows about it's facade it's going to use. It just abstracts that information away, the python code has no such idea.
[17:02] <nammn_de1> stickupkid: thanks and dammnit
[17:03] <stickupkid> nammn_de1, see https://github.com/juju/juju/blob/6ea20273f7c540d93b1075d77aafe39e70770b33/api/controller/controller.go#L36
[17:03] <nammn_de1> stickupkid: doesnt this suck in this case? Would need to add a new facade for this to work in python, right?
[17:04] <nammn_de1> ah
[17:04] <stickupkid> nammn_de1, nope, we export all facades for python to consume
[17:04] <nammn_de1> stickupkid: i need to understand how the attributes are set in python
[17:05] <stickupkid> nammn_de1, there is a set of facades we don't export https://github.com/juju/python-libjuju/blob/master/juju/client/facade.py#L21
[17:05] <stickupkid> nammn_de1, maybe it's that one
[17:28] <nammn_de1> stickupkid: this quick hack did it for me: https://pastebin.canonical.com/p/sBskSH7QdQ/
[17:30] <stickupkid> nammn_de1, just remove "ClientFacade" from this https://github.com/juju/python-libjuju/blob/master/juju/client/facade.py#L21 and then follow this guide https://discourse.jujucharms.com/t/python-libjuju/1553
[17:32] <nammn_de1> stickupkid: I will keep that in mind,ta
[22:35] <timClicks> babbageclunk: ping, question in discourse for your perhaps? https://discourse.jujucharms.com/t/application-level-data-questions/2447
[22:36] <babbageclunk> timClicks: thanks! I'll have a look in a bit