[00:58] <MarkMaglana> hi jam when you're online and have time, i'd like to get some advice on something i wanted to do with interfaces. ultimately what i'd like to achieve is for the interface to have zero knowledge about the charm's domain. rather, i'm trying to just make it provide a data structure to the providing charm and let the latter populate that with the right data which the interface then shoots over to other side.
[05:23] <jam> morning all
[05:49] <t0mb0> hi jam I keep getting application-wordpress-k8s: 05:46:42 DEBUG unit.wordpress-k8s/30.start ops.model.ModelError: b'ERROR json: unknown field "readinessProbe"\n' https://pastebin.canonical.com/p/JDc8tjPMfs/
[05:49] <t0mb0> when I use pod-set-spec version 2
[05:49] <t0mb0> works for version 1
[05:50] <jam> t0mb0, I'm not aware of the exact differences, but I do know that some fields are moved around for pod spec v2
[05:51] <jam> t0mb0, https://discourse.juju.is/t/updated-podspec-yaml-new-features/2124
[05:51] <jam> t0mb0, looks like in v2 the 'readinessProbe' goes under a "kubernetes" section
[05:52] <t0mb0> ah okay cool thanks!
[08:04] <mup_> Issue operator#295 opened: Add resources to harness for oci-images <Created by chris-sanders> <https://github.com/canonical/operator/issues/295>
[08:04] <mup_> PR operator#296 opened: Oci resources <Created by chris-sanders> <https://github.com/canonical/operator/pull/296>
[08:51] <Chipaca> good (belated) morning all!
[09:02] <jam> morning Chipaca
[09:07] <Chipaca> jam: how was the break?
[09:08] <jam> pretty good. mostly stayed home. did go out for lunch one day
[09:09] <Chipaca> jam: oooh! like in the before times!
[09:09] <jam> indeed. they had the tables spread out, and the menu was a QR code to a PDF so they didn't hand out paper to people
[09:10] <Chipaca> so like the before times, but updated for the dystopia we actually live in
[09:10] <Chipaca> fair
[09:44] <jam> Chipaca, we ended up playing cards and Risk for one day, took several hours to get close to an end of Risk, it was actually pretty fun and back-and-forth.
[09:44] <jam> How was your extended holiday?
[10:02] <Chipaca> jam: not bad! didn't do much at all, which was perfect
[10:02] <Chipaca> facubatista: charmcraft seems to be needing quite a bit of love :-(
[10:03] <Chipaca> jam: any chance for a review of ops/lib ?
[10:11] <jam> Chipaca, sure. is that #244, or is there a different PR?
[10:11] <mup_> PR #244: ops library support <Created by niemeyer> <https://github.com/canonical/operator/pull/244>
[10:11] <Chipaca> jam: that's the one
[10:17] <jam> Chipaca, looking
[10:44] <facubatista> Muy buenos días a todos!
[10:46] <Chipaca> facubatista: buen día caeza!
[10:46] <facubatista> hola Chipaca
[10:46] <facubatista> Chipaca, thanks for the review
[10:57] <facubatista> Chipaca, symlinks are not zipped as links, but as the "copied file"
[10:57] <Chipaca> facubatista: the way we're doing it yes, but it's not the only way
[10:57] <Chipaca> you can have them as symlinks
[10:57] <Chipaca> just needs a bit more dancin'
[10:58] <Chipaca> facubatista: you can see it with the 'zip' tool if you want, 'zip --symlinks' exists
[10:58] <Chipaca> in python zipfile you need to twiddle things a bit but it works
[10:58] <Chipaca> facubatista: https://stackoverflow.com/a/60691331/348012
[10:59] <facubatista> Chipaca, we need to define what stuff in the build directory be like, for "while developing" behaviour; then we can use zip to support what we did there
[10:59] <facubatista> Chipaca, ack
[10:59]  * Chipaca starts warming up the valves on his webcam
[11:00] <jam> brt, my son has a homework questi
[11:31] <Chipaca> facubatista: charmcraft's setup.py is missing some magic pixie dust or something; it isn't pulling in commands/
[11:35] <facubatista> Chipaca, oh, nice catch, will check, thanks
[11:40] <niemeyer> mup_ is now in a better home.. please let me know if something bad happens with it
[11:41] <niemeyer> I'll be moving the remaining of the configuration from mup to mup_ in the afternoon, and hopefully killing the old instance
[11:41] <Chipaca> niemeyer: ack
[11:52] <facubatista> niemeyer, thanks!!
[11:58] <Chipaca> facubatista, jam, i'm going to be 5 minutes late
[11:58] <jam> ok
[11:59] <facubatista> Chipaca, ack
[12:00]  * facubatista fights auth
[12:01] <niemeyer> facubatista: np!
[12:11] <mup_> Issue operator#295 closed: Add resources to harness for oci-images <Created by chris-sanders> <Closed by jameinel> <https://github.com/canonical/operator/issues/295>
[12:58] <Chipaca> coffee?
[12:58] <Chipaca> mup_: coffee please. Preferably a flat white, but otherwise a cortado would be fine
[12:58] <mup_> Chipaca: I apologize, but I'm pretty strict about only responding to known commands.
[14:04] <mup_> Issue operator#297 opened: Remove default target for Framework.observe <Created by jameinel> <https://github.com/canonical/operator/issues/297>
[14:10]  * Chipaca does the changes and sees a barrage of EEEEs
[14:23]  * facubatista brb
[14:28] <Chipaca> jam: would you say that StoredStateData's on_commit is indeed a public api?
[14:37] <mup_> PR operator#298 opened: ops.framework: Remove default target for Framework.observe <Created by chipaca> <https://github.com/canonical/operator/pull/298>
[14:47] <jam> Chipaca, offhand I wouldn't think it was meant to be public to others
[15:11] <facubatista> jam, Chipaca, I included the fix in setup.py in https://github.com/canonical/charmcraft/pull/9
[15:11] <mup_> PR charmcraft#9: Kill bogus commands, and logged the error when happened <Created by facundobatista> <https://github.com/canonical/charmcraft/pull/9>
[15:30] <knkski> What is the state of the art for including dependencies in a charmtech charm?
[15:31] <knkski> previously, it was git submodules, but i'm now seeing references to env/
[15:31] <knkski> and i'm wondering how that works exactly. am i supposed to vendor all depenencies?
[15:32] <Chipaca> knkski: where are you seeing references to env?
[15:33] <knkski> Chipaca: from the readme on https://github.com/canonical/operator:
[15:33] <knkski> ├── env/│   └── # ... your Python dependencies, including the operator framework itself
[15:34] <Chipaca> knkski: depending on what you mean by vendor, git submodules is that already isn't it?
[15:35] <knkski> Chipaca: i also see "and any Python dependencies (maybe using some kind of virtualenv)"
[15:35] <Chipaca> knkski: so, we are no longer recommending git submodules. Which does _not_ mean we are dis-recommending them :-)
[15:35] <Chipaca> knkski: if that works for you, feel free. If you'd rather have a requirements.txt as you would in any python thing, do that
[15:36] <Chipaca> knkski: does this make sense?
[15:36] <knkski> Chipaca: so if i have a requirements.txt, how exactly does that work as far as actually getting the items in it installed?
[15:37] <Chipaca> knkski: can I answer you as if it was this Friday?
[15:37] <knkski> Chipaca: sure
[15:37] <Chipaca> knkski: or do you need my answer to be correct and true _today_ :-)
[15:37] <knkski> Chipaca: lol, friday is fine
[15:38] <Chipaca> knkski: 'charmcraft build' defaults to picking up a requirements.txt if it exists, and use pip to bundle those into a .charm file
[15:39] <knkski> Chipaca: how do i get charmcraft installed?
[15:39] <Chipaca> knkski: pip install charmcraft (this answer is also future-dated; until then github.com/canonical/charmcraft )
[15:39] <cory_fu> Chipaca: Has the question about different architectures and Python versions been sorted out, then?  (Though that only applies to dependencies which require compilation, like cffi, though I think PyYAML has optional compilation so might hit that as well.)
[15:40] <knkski> Chipaca: thanks, i'll try it out
[15:40] <Chipaca> cory_fu: charms will have a 'platform' which is like snaps' 'architecture' but richer (this is not there yet so for now we're ignoring it)
[15:41] <cory_fu> Chipaca: Ok, that sounds great.  I look forward to that.
[15:44]  * facubatista missed all this conversation :(
[15:44]  * facubatista brb
[16:09]  * facubatista is back
[16:33] <Chipaca> facubatista: anything I can do for you right now?
[16:33] <Chipaca> otherwise I'll go for a run and check in later
[16:33] <facubatista> Chipaca, nop, thanks1
[16:55]  * Chipaca goes
[17:00] <knkski> Chipaca: ran into an issue installing charmcraft, and opened up a PR: https://github.com/canonical/charmcraft/pull/10
[17:00] <mup> PR charmcraft#10: Fix pip installation <Created by knkski> <https://github.com/canonical/charmcraft/pull/10>
[18:22] <Chipaca> facubatista: what do you think of knkski's fix, compared to yours?
[18:22]  * facubatista cjecks
[18:25] <facubatista> Chipaca, the "packages" option is simpler in my case, don't know why the second change is needed
[18:26] <Chipaca> facubatista: if theirs does what I think it does, it's more future proof, no?
[18:27] <facubatista> Chipaca, it looks that if in the future we add other subdirs in the project they will be automatically included
[18:27] <facubatista> Chipaca, not really sure if that's a good thing or not
[18:27] <facubatista> Chipaca, don't know if the second change is related to the first one, also
[18:35] <Chipaca> facubatista: I'd have to learn this stuff to answer :
[18:44] <facubatista> Chipaca, I'm ok with either solution, as long it works
[21:35] <Chipaca> facubatista: reading the docs, their console_scripts seems more correct, and the find_namespace_packages looks to be the right approach imo (because of the include= arg)
[21:51] <facubatista> Chipaca, let's land it, and I'll merge my cleanup branch after that
[21:51] <Chipaca> facubatista: sgtm
[21:56] <facubatista> Chipaca, landed it, and then merged master in my branch and pushed it...
[22:22] <Chipaca> facubatista: neat
[22:48] <mup> PR operator#299 opened: charm: make role an enum, with 'peer' instead of 'peers' <Created by chipaca> <https://github.com/canonical/operator/pull/299>
[23:01] <mup> PR operator#300 opened: charm: use magic strings a little less <Created by chipaca> <https://github.com/canonical/operator/pull/300>
[23:02] <Chipaca> ok, i'm off to bed
[23:07] <Chipaca> hahaha i lied
[23:08] <Chipaca> ok, now yes
[23:08] <mup> PR operator#301 opened: charm: role is 'peer', not 'peers' <Created by chipaca> <https://github.com/canonical/operator/pull/301>
[23:09] <Chipaca> 299, 300, and 301 are all fixes for the same issue. Let's see which one wins the peer review dance :-)