Kubernetes API Expression Working Group, 10 Nov 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: WG API Expression Bi-Weekly Meeting for 20201110

Description

WG API Expression Bi-Weekly Meeting for 20201110

A

All right, hello and welcome to open group api expression, it's the 10th of november 2020, and we do not have a topic on your agenda right now. So if you have something feel free to bring it up.

A

Silence so I'm switching to the topics.

A

All right, I think we haven't, talked much more about the gradation criteria.

A

Yeah. No, we haven't. We are approaching code freeze this week, so um I think the um idea that we would make it 121 it's a bit more ambitious than thought previously. I don't.

B

Know what makes you say that.

A

That it's already 120. I thought we had more time honestly, but it's december or something.

C

B

um No, I think, yeah we're focusing on this uh 120 item right now.

A

B

And um which we are fairly confident is gonna go in and then I'm I'm expecting we're going to focus on the uh ga graduation during the next cycle.

A

Yeah that sounds good.

A

It just will. Next code freeze is for 120 right, yeah, okay, yeah yeah, um anything new in the built-in declarative default.

B

Lots of new things- um this is updating defaulting code. This is jeff is jeff is working on that. Do you want to talk about it.

D

Jeff, oh yeah, sure um hello, everyone nice to meet you, I'm jeffrey, so yeah. I have a pr out for updating the uh default code generator um to incorporate antoine's changes from the defaulting check. It should be ready to review now.

C

B

Joe yeah, so this is good. It's going on um what is next yeah, so that's merged.

B

We have the conversion of the open api schema into the smd schema.

B

This is uh under review by stefan. I think we I'm expecting hoping he's going to review today and give an lg gtm, um but he's been not reviewing for the last two days and thus well really in a rush right now. So I'm probably gonna stop looking for another viewer for these um and then these implement the new crd cement. Oh that's great! That actually merged.

B

um It's more than started, and this is merged so basically, what's left, for this is the defaulting code from uh jeff and this pr, which I think is very very close to munch and and so yeah I mean. Ideally, we watch this one and we get a review for this one and we have two days to retie it on that which I think is.

B

I mean we also need to integrate things in kubernetes, but um jeff has the pl that does that already and we can see that everything works as expected. So um I'm confident we can do it.

A

Awesome, what about the smd part? Oh.

E

B

A

B

Yeah we have to bump smg cube open api as we merge the spl and then gen uh the gengo. However, you pronounce that uh we need to merge all of these and then update the default in kk, which uh jeff has already done. You know work in pogo, spr.

A

All right sounds very good. A lot of progress.

B

Yeah, this is the same thing.

A

uh Status, wiping yeah I I made progress with what we discussed and um the the implementation is smd seems to work as I expected to be. I wrote a few tests. I want some more tests to be sure.

A

Also right now I hard coded the pod strategy into field manager just to make just to test it and going to the type value seems to work as well. So now I'm gonna um get in my smd changes and try it for real, and if that works, I'm gonna start wiring the strategies into the field manager or the prepare for update and prepare for create, and if that works, uh it might actually.

E

Be done for once wow, that's exciting! How come are.

A

um The smd test.

B

A

um The the function I I I did now would just interact with the managed fields and the live and object after prepare for create, and it's not that much to do it's quite simple to to just do it if here and an intersect there, and so the test just compares managed fields before and after, and it makes sense. So I'm just I don't know. What's the right word in english nervous, anxious unsafe that, because it looks so easy that it might be wrong, but.

F

We would have expected something easy out of this yeah I mean.

A

After all the time it took, I can't believe making progress. I don't know cool.

B

I've set up a reminder to look at it in a good week. I think, because we're gonna have the code freeze and then the documentation deadline and.

C

B

I look at it right after.

A

Yeah, even if I would finish what I could, because I'm I'm almost approaching my holiday or my overtime is done now, so I don't have to work anymore for the next month.

A

um I could finish it to covers, but I don't expect to get somebody to review to change with wiring and everything in that short amount of time, so cool. We have something very early for 121. Then.

B

That's good, that's very good news, because that's that's important for uh the ga graduation and the audio.

C

B

Do something else.

C

Yeah cool, thank you.

A

All right tracking sub resources.

G

uh Yep, um the pr is there uh julian review it thanks. um I've I've updated some of the um comments there, so yeah um it's ongoing. If anybody wants to have another look.

G

If the approach in this pr works, um the prs for the rest of the resources should be fairly easy, so yeah.

H

Okay, because you're saying this is what do you mean by that.

G

This is just for a scale in the deployment resource, and then we can, we can do it for replica sets and store, and um crds and etc.

B

Okay, that sounds good.

A

I'm gonna look at it as well. Thanks all right, dangling fields are blocking.

B

We have a bug and we haven't spent the time to look at it yet, and this is a bug we'll have to back parts into maybe 116, but certainly 118 17.

C

um Do we have any deadlines for backboarding those or.

B

I mean we don't have a fix so.

A

Okay, yeah well yeah.

E

B

um It's maybe a good idea for someone who wants to try and know more about smd about the merging code and stuff.

A

B

A

um I I I found out that yeah.

B

But it's it's maybe doable.

A

Yeah um I'll I'll see how I continue with the status wiping and I could look at it. I spent some time with it, but if somebody else wants to check out.

B

A

That's maybe even smarter.

B

A

Or we can pair on something I don't mind.

B

Yeah, I don't expect, I don't yeah somebody who ends up understanding. This code could probably fix it in a week, uh but that's probably two weeks of work, maybe to even just understanding. Maybe a week. I'd say two weeks. I'd say: that's probably two weeks of work.

A

uh It has a cube ctl.

C

A

I

You're unmuted but you're not talking. Oh sorry, my mic was muted.

I

Yeah I mean I think I need to take a look at this. I I think we, this looked pretty good the last time um I saw it, um I'm not sure, what's not what's next to do.

B

Yeah, that's my perspective too. Did you have a.

A

Problem with the sea life people.

B

Oh, that's a good event. That's a good idea, um let's schedule a meeting for that, I'm not sure what we meant by that. um Oh, I know. Okay, um I'm gonna replace that by let's start qr6 cli about it after 120.3.

D

A

B

Yeah joe's not here, but he gave me a status on.

B

Let that look it up.

H

He went to the.

J

Prototype but it doesn't use the right type of code generator to put in to um queue builder. So that's basically the status right now.

B

Okay, yeah, he told me, went to the builder uh meeting and they they have people who are going to trade out and um and see if it's good or not,.

A

All right, yeah, nothing new here. This is documenting.

B

That's for next week, if uh we should probably, as could freeze, is gonna hit. We should synchronize on the channel and see who wants to do what, with documenting.

I

B

But ideally, if we can spend the week trying to carve off these issues, that'd be nice.

A

um Is that one done with the pr above or was it just open api to smd.

B

Jenny's gonna comment on that.

J

um Sorry, which I don't know that structural schema to smds yeah. um I have a pr for this still, but I'm it's going to work on testing it. I think, and I have to rebase it onto the new like changes.

A

Okay, I I asked because it sounded very familiar to you: open api schema transfer, open api schema to smd schema.

B

Yeah, so basically we have a layer of interaction between the open api and the smd. So if you look at my vr, you have to change this layer in between we convert to another type, then convert back to smd we'd like to skip this test.

B

The intermediary step does not capture the everything we need to capture.

A

A

Okay, we still need to investigate that.

A

Empty pot select doesn't mean empty.

B

Oh interesting so yeah this one's done, but not cherry picked.

B

Oh that's the status on this. I was looking at the status for that yesterday, um yeah. We need some people to help cherry picking that.

A

Was this again that we, the the one where we had to pick into the api as well, the where the topic was a bit more complicated.

B

Not sure I mean we yeah, we need to try to pick the smg version, the smd code on all the smd branches and then jpg the smp bunches on all of the release, bunches of kubernetes.

A

Could use a script for that.

B

F

Yeah I mean yeah, I remember the last time we tried to do that and it was a mess yeah. I can help.

B

Someone and the one below yeah we have someone trying to cherry pick. These.

D

B

And but the 117 is slightly more complicated because it conflicts ga sold. That's yeah, yeah, yeah yeah. It is um just needs to be checked.

A

Okay and uh 119 118 is waiting for review.

B

This one, I don't think we can fix, and this one there is a fix, but stefan disagrees with the fix.

A

Yeah right, the comment I mean that makes sense. It was about also being float right.

B

B

Yeah, should we bring the things in the in the backlog that are blocking the ga.

B

There are not many just behind the api user documentation. Okay, I'm.

D

B

It up into these, so we've don't forget to do it. Yeah. That's.

A

That's the only one I can see right.

B

I don't have many ga blocking items, so I'm that's. Why I'm confident we can do that in the next cycle.

B

uh Good, I'm good I'm going to uh maybe uh let leila introduce herself since uh she's new to this group.

A

K

Yeah hi everyone, I'm layla and I'm happy to see all the progress still trying to catch with the documents and the things that needs to be done for the ga, but I will be happy to help so I don't know the details, but uh I will try to learn and contribute. So thank me at any time if you need anything from me and that's all yeah. Thank you.

B

What kind of help you wanna uh that you wanna bring.

K

uh So it is uh first of all in the planning that I am going to see all the documents and see what we can do there and then for the code review for any issues that needs help and, um let's see but whatever, if you need to think about anything, brainstorm uh go through some ambiguities, anything that you uh think of I um so I'm new to everything. So you can, um you can select whatever you need, yeah.

E

K

And just add me to that.

B

Thank you later.

K

Thank you awesome. Welcome.

A

Just do we have any any tasks. That would be great starters that you would be interested.

B

B

Yeah and think about it.

K

K

B

I mean we have things about the topology, uh that male love.

B

So that's kind of uh that's. Okay,.

K

B

That's probably a good topic. I.

K

Will take a look at that.

B

Yeah, let me know if you need.

B

Anybody want to say something.

A

Else doesn't sound like it, except of stay, safe and sane. As usual,.

E

Yeah, if possible,.

B

Thank you so much. Everyone see.

G

B

G

J

Thank you bye-bye. Thank you. Bye.

C