Cloud Native Buildpacks Platform Sync, 28 Apr 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Platform: Sync 2021-04-28

Description

Meeting notes: https://bit.ly/38pal2Z

A

All right well well, well yeah, another platform, sync, uh I guess I can get things started with status updates. um So I guessed there are a couple releases or a couple pr's floating around for all of the asset cash stuff. That needs to be reviewed by couple different groups of people, um but I think waiting kind of just waiting on hot air to come back. So we can start looking.

B

A

That and I'll probably like try and give some sort of like demo or something like that about how it all works. So it's not just like a big block of code um other than that been trying to kind of um talk to the paquetto people a little bit maybe try and get some more involvement from them, especially since, like a couple of the conversations, I've had, especially the one with patrick at space force, kind of was around or sorry spaceship.

A

uh It kind of seems like that. Like final last step of allowing new users to kind of onboard and use things effectively, slash figure out how build packs work together, it's really the stuff we're falling short on or could improved. So that's pretty much. It.

C

Sweet um for me, I know I've been doing a little bit of work on modifying the image feature uh of the back code. It's pretty you know I. I did talk to heavier about this, but you know it's pretty straightforward. When I'm trying to go towards it's.

C

Basically, if you have the daemon um use, it basically just just use it, and if you don't have it try to pull a remote image and you know whatever errors you get right and the reason we're doing it is because the daemon can already pull remote like it can already handle more edge cases than we can. So it's better to just throw that to damon and return whatever error it gives, which would uh avoid the issue that this part anyway.

C

The implementation is there and I'm pretty solid and people want to take a look. It's not too big, I'm just having a lot of trouble with the tests right.

A

C

I've always had with with this repo, which uh not a big concern, but you know I was thinking you know it would be really nice to do something about it. Like you know, something I'm thinking is just like.

C

First is, you can have like one block of test run like unit test run first and then before having everything else run. You know it's just so it's easier to parse failures. Basically I I guess just because I'm speaking out loud, I'm dragging out, I guess I'm just trying to say, like it's very hard to parse a failure.

C

uh When you push up the ci, I guess and hard to debug too right.

A

Are you getting acceptance test failures.

C

uh So yesterday I was getting unit test failures, which of course is my fault right, yeah. Of course I'm screwing up, I'm not taking that, and I think today it's acceptance test failures which I haven't I haven't run luckily, but.

A

Okay, I mean I'd if you would like to pair on this, I feel like or like I feel like, just by figuring out what things to search for in that tiny little github window on the acceptance tests. You can narrow down what you're looking for and how to find it. But if you don't know the exact strings to search for it's just like 50 pages of blah, so that's probably something we should document, very least so that other people don't have the same problems yeah. I think javier showed me this, but.

B

I wonder if, given the um given that those test outputs are available from a url, I wonder if even like a little utility or something like that could be helpful, like grab the most recent failures and then I don't know, maybe that's the.

C

B

Or I mean maybe it's what you're suggesting to anthony just like change the way the test output renders and make it easier to purse. To begin with, I don't.

A

Know yeah. I have no idea how to do this, especially.

C

A

Actions where it's just like here's, your block of output, enjoy great.

B

No rich go options or anything.

B

A

Okay, yeah! That's that's a good idea, though,.

C

It's good yeah, I'm I'm still thinking about it. I'm still thinking about it. I'm sure. There's a you know. Nice button-up streamline solution here somewhere, but I just.

C

Don't know it yet.

B

Yeah no I'll throw in there too. It's not at all a help, but a predictive help of if you do need to recreate an environment very similar to the um to the runner, the linux runner. I had to do that last week trying to fix the ci failures um due to the docker version change.

B

So if you need that, I can send that along just like the identical azure image that the runners made from and um docker versions and things like that commands to install because they use a special azure, specific version of docker, which is um close to what's uh upstream. But you just have to add the right app sources for.

A

A

Yeah, I also feel like anyone who's pushing like outside of this project, pushing a committee for the first time and getting like a failure on a windows machine or something like the lips gonna be like yeah.

A

No, there's just no way, there's so many layers.

A

Yeah everything think about that: okay, um yeah, I'm just gonna- add some notes here.

A

C

All right all right, we did say we're gonna, be a little bit better about these notes. So that's why I'm trying to bolster these a little.

C

A

A

Just I think this is pretty easy just to add some like.

A

And surround this stuff until we have like an actual good solution that people are happy with.

A

Okay, cool, I guess, uh release planning. We have a release slated to go out in just over two weeks.

A

I have a lot of things in the next milestone, but I think I'll have some time to actually work on those now that most of the like asset package, stuff is kind of off my plate, but in the like ping pong match of waiting for other people and then delivering a little bits of feedback, so but some of this stuff might have to go.

A

um I just talked about talks heavier about that, because I don't think I have the permissions required to move anything around.

A

Actually, let me share my screen really quickly see how much stuff we have in here.

A

I think it's a lot.

A

A

Okay, actually we're doing pretty good.

A

All right, we got some stuff.

A

Juan said that he had kind of come to a conclusion on this. Actually.

A

I think this is that right, yeah, let's just let's mark him as the order of this for now.

A

Okay, yeah, I don't know if we're gonna get through all this stuff.

A

A

All right, uh the next thing is needs discussion. There is only one thing: it is waiting on el bandito aka travis, to give us some help here, since it doesn't it's not immediately clear what he is asking for here.

A

That's pretty easy.

A

And finally got some rc's: okay, oh yeah this. So this came up the working group last week and I'm just I think one of the questions that emily raised was around what exactly we're approving here. I'm like.

A

I'm super happy to just to prove this in like spirit because it's the right direction, but I also like people are like a little bit inclined to like jump into the like nitty gritty details.

A

um I don't know if there's any conclusion about on that conversation, the working group anthony. Do you remember.

C

Yeah um I was present and I also didn't get any clear outcomes right I mean to me there's a couple options. Let me to me is get these details out into into individual issues right like if this rc is approved, which I think is mostly on here.

C

Maybe we can have more better details and issues which which I'm happy to drive out right. I've even put some of the issues I'm thinking about um another one is: maybe we can have a discussion thread about it about the details right. I think I think the problem is really just the details which, which haven't been spelled out right. So I'm just I'm just wondering what is the best way to do that and I think everybody's aligned on moving forward with the rfc without the details being being spelled out as they are right now right.

A

Okay, all right so can we just like have some, because I think that that was like a similar conclusion is if we try and plan everything out. This is going to take way too long, but everyone is definitely excited about like everyone's like okay. This is definitely some functionality. That'd be useful.

A

A

A

Actually, I don't know if I'm gonna want to freehand this entire comment, but.

A

And I almost wonder if, like just.

A

A

I'll go back a little bit.

A

Does that seem like reasonable? I just don't want this to get stuck in like the bike shed phase, because there's so much possible stuff cool awesome, that's good to me!

A

C

A

C

Where my thoughts were at right last time in this previous sync, I'd asked, you know when when do because I'm new right, when do the issues usually get made, and they said it's usually when it's in the final comment period, so I'm waiting for it to go to the final comment period. So I can spell out the issues and you know we can we can.

C

I don't know you know dissect.

A

C

There and and then just telling you where my head is at yeah.

A

Yeah we had like another rfc that was kind of like this too right around refactoring, where it's like. No we're not going to make any of the conclusions about what exactly the final package structure should be, because that's just like not a good use of anyone's time should wait like defer on the details.

A

Just like making sure people understand that.

A

Okay, cool that that seems good. Well, hopefully, it gets uh fcp soon cause that looks super sweet. um I think that's everything on the agenda.

A

Wait, no did I miss an rfc.

A

I totally did sorry, that's two pack cash options. um Oh I'm supposed to review this. I think it looks pretty good.

A

Last thing that I remember reading is that there are some questions about whether we care about having the launch cache. I don't think we should.

A

Oh okay, we'll wait for out here to come back. Let's work, his thoughts on that and lastly, this looks like a draft.

A

A

Any last-minute topics.

A

No dang, okay, that's.

A