DFFML Weekly Sync, 13 Apr 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Weekly Sync 2021 04 13 Part 2

Description

Meeting Minutes: https://docs.google.com/document/d/16u9Tev3O0CcUDe2nfikHmrO3Xnd4ASJ45myFgQLpvzM/edit#heading=h.i74nv6ns2nz8

A

A

um So this is what we talked about where there's something going on where, um where these tests were written to assume that they were running relative to the to the repo root and.

A

uh Yeah, so we need to just figure out where I mean we need to fix these test cases. For this to, for this to work right, um did you have you looked into? Have you tried to debug this yeah I've tried but didn't find anything. So, okay, we can try.

B

To debug it all right: okay,.

B

I thought I opened this already and I think, like gbm pr is ready to home. I mean there's nothing in this.

B

A

Oh, that's right. I have this demo great.

C

A

Okay, so right, so what? What did you? What what have you tried so far to to work on this issue? Because we can start from there the async test case, one.

B

Oh, I just changed into the relative path right yeah and after after that, I have tried to solve this thing, but uh didn't come up with the solution, so didn't have any.

A

Steps right so and then this one, I think the.

A

The main thing here was just the screen used to be. um I think I I somehow my my changes got lost when I added that other example.

A

So and we really need to fix that white space check. um Oh, I think we were waiting on the merging the accuracy stuff.

D

Have we created that list of points that you may have.

A

The list of why.

D

A list of points everyone wants to talk about is that thing done.

A

uh No we're still we're still gonna do that. I just wanted to hit a few things quickly. First, um okay, let's see this guy, because these are things that it thinks things that the ci is going to run is what my where my brain is going right. Now, um let's see where did that go model?

A

Where did that go?

A

A

A

Yeah that way, we kick off the ci on a few of these guys, uh while we talk.

A

Okay, so um let's just run the talks.

A

A

A

Yeah now we can run that.

A

A

Okay, so let's jump over here- and we can start structuring this now that we have a few things- kicked off.

A

A

All right so shaw.

E

uh Yeah, uh I made the changes to the data from pr as requested, and uh should I make uh I want to discuss if I should make a pull request for the forecasting what models thing just yet, because that uses a second party plugin. So should I hold off on that or okay.

A

uh You know, I think I think so I talked to I talked to to yash and- and he said basically that he's gonna have a little more time um upcoming. He he didn't have you know, he's had exams and things, so he hadn't had time yet to get to that. um So we're gonna go ahead and we're gonna try to merge some of the second party stuff in um just because you know for for the selection period of gsoc, like the contributions will will count up until we we decide right.

A

So um so you will, you know we just we'll just try to get those merged, and then you know it's very clear that those contributions are. You know um you know what they should be right. They they're they're countable, um so we'll yeah try to make another pr and, and then that will that will account for that so um data pr video for review will.

E

A

um Forecasting model pr um and I'll make a note of this. Actually, in that.

A

General and general business.

A

It's like general business is a person attending our meeting. uh Okay, so, let's see um second party plugin are happening. If you, our time schedule has.

A

Changed so the time schedule for the second party plug-in stuff has changed basically make sure we're still recording.

A

um So we're going to do that, you know we're targeting like two or three remember. You know two to four weeks out to be complete with that.

A

uh Well, probably three to four, because starting two weeks now breaks out to complete. um So you know the impact of this is.

A

That we're going to try to merge um all uh prs related to second party or you know all plug-in prs.

A

And then so, gsoc wise.

A

Let's look at the rubric or whatever.

A

Yeah contribution so um like yeah, I think this is basically okay. Maybe it doesn't say it straight out, but basically you know up until the point this is like we could we cut off like right right before so, if you're still making contributions, you're you're getting points in that category um so and that's merged pull requests um so jsoc wise.

A

uh If you yeah, you need more contribution.

A

Points. um Try to get your try to get your plugins merged to a mergeable state.

A

Okay, so and then you'll make the forecasting pr anything else from yusha uh nope all right, okay, so and then so how's that linting how's the linting one going and we finished the doc support great.

D

A

D

Yes, yes, uh so uh I in landing thing I was working on those uh I was checking like how much percent is failing. So I saw like 30 35 percent of the commits in master uh up to this point are failing all right uh in our language.

D

So so I I just started grouping them by you know type, and I have found some uh body mutations, which I have implemented some body mutations, but haven't pushed them because it is very complicated at the point all right.

A

At this point, I want to make it more.

D

Clear great yes, I'm working I'm working on that, so uh we will not be able to touch 100 offices.

A

Yeah yeah yeah, that's.

D

A great approach, so that's what I'm doing and uh I had a few questions other than that, because you know I'm having some tests this week. So I was not able to write much code.

D

One is about uh 1078 uh issue, number 1078, okay, um so uh issue 1078. If you've been like, uh maybe okay,.

D

The question is that uh we had a word on uh making that async. I o sub process one wrapper for, like you know, we were usually using similar code everywhere. Yeah.

A

D

Wanted to make it a single function, but uh actually the thing is uh I I think we have something similar already there in uh testing console test command start by so should oh.

F

D

Extend its functionality without writing a new module, or we should write a new module.

A

D

And uh like a subprocess can written a lot of things like written code and standard out standard error. So how should the user be uh giving input to that function that you know like how the interface should be yeah.

A

D

So that was one of the questions. Another one was uh like I'm not very sure if we can exactly calculate the run time of async functions, but I think having a time decorator would can be helpful. Can it be.

A

Like a time, okay, so you're you're, saying like in general for for um like a general time decorator for logging or something.

D

Grab that as well.

A

Hey, that's a great yeah. Do it that sounds great 1088. Okay,.

D

um So we can like exactly see where things might be going. Yeah performing.

C

Or something like that.

A

um Yeah, so uh I don't know if we have a dfml util log, but that sounds like where that would go right.

D

D

uh Measuring thing we didn't have something.

A

Okay, I'm just saying if I don't know if we have a dfo from all util log, but that sounds like where that would go um like that would be a good place for it. If you're gonna put a new a new thing in should.

D

Probably okay and one another thing was like. Sometimes if I see certain things being added by contributors, and I know that something is wrong, something is right. Can I comment on it? Oh.

A

Yeah definitely yeah no, so this is yeah if, if, if, if you guys see other people's, you know pull requests or or stuff that you you you, you know you, you feel that that you have some input on. You know you can feel free to comment whether whether you're sure you're wrong or whether you're sure you're right or whether you're you're wrong it doesn't matter. You know the point: is it's good to get discussion right like that's? What a lot of what we do in this meeting right is.

A

This is sort of a central place for us to for us to discuss things right and if somebody realizes you know in this meeting as we're talking about something, then we can. All you know raise raise, concerns right and figure out what the right approach is. As you know, there's really you know, there's there's no.

A

I mean there's, there's a right approach and a wrong approach from a functional aspect when you're programming right but there's no, you know any approach that gets the job done, is sort of you know kind of the right approach right, but we're we're looking usually for the more you know. The more like you know, the more maintainable and elegant approach is usually what we're going for right, understandable and maintainable, and then you know elegant after that right and then elegant bean.

D

Very, very obvious that we can like really point out like yes,.

C

D

Wrong, this is going right, so at least we can tell the contributor that, yes,.

A

D

A

Yeah helping each other on that type of stuff is is, is very important. um Actually, I think that's one of the rubric um recommended rubric items too. It's like you know helping you know, helping helping other people, uh you know with their prs and stuff. I think that was on there. I think, uh let's see uh where is it.

A

Yeah engagement with community right, so you know answering answering other people's questions, pull requests right. You know this is this is this is this is how we help each other right. um So that's! That's a very you know, that's a very important thing to do um so. Yes, if you, if you see something, then definitely uh provide input right, you know, don't just provide employers. The counterpoint to this is don't just say something just to say something, because that doesn't help anyone right.

C

You know if you're, if.

A

You're homeless, right.

D

A

Right, I think that's what you're saying so so yes, you, you hit the nail on head: um okay, um okay, so buck appears by um or no bucketed bucketed commit messages by.

D

So so there are two main mutations that I'm looking into yeah uh one one is like the the last part contains a function or a heading, something like that, which is there in that folder in a file, so that would be searched in a specific subfolder and and another mutation was actually I've, just missed it. There was another mutation of making capital and making small like someone that does us, like you know that is going to be a thing like yeah folders are capital names, and some files are capitalized completely yeah.

A

D

Don't write that in commit messages, yeah.

A

That's actually right.

D

That is correct, so that these are the two things I think that will solve a lot of stuff, and even after that, we will be able to clear even more.

A

All right, okay, yeah and those those sound like you know rather important ones. To me, I we talked a little bit about the function. One last time I think- and we sort of said- maybe maybe it would be optional, um but if you're already there and looking at it, then then maybe it makes sense to do and we can always expand on it later. I think the capitalization one.

A

This seems like something that we want to get right on the first try because there's a lot of you know we want to make sure this mutation rule works before we merge the pr, uh because there's there's that's a frequent issue that we're looking at um and then, let's see, abstractions around the concept. That's already existing console test. Okay, so then the response on this is so console tests.

A

um Actually I did a little. I think you guys have all seen console tests, but I did a little demo video that I'm trying to post to youtube. If I can figure that out. I forgot about that, but um this is actually going to get split out into its own thing. Within this is console test will get split into its own package within the second party plug-in org, because it sort of doesn't need.

A

You know it doesn't really need to be tied to dffml um and so we're not we're going to try to not cross cross the term people uses cross pollinate but, like you know, um use things in one that are in the other right. um So, uh let's stay away from stay away from using um using dfml stuff in console test and using console stuff, indeed or well.

A

We can use console test stuff in dffml, but I think that the one that you're talking about, maybe is not the best suited for this, um because I think we, I think, let's see which one is that.

C

The parents would be independent of the children right. Is it like that.

A

D

Children are surprised.

A

Yeah yeah yeah, the parents should be independent of yes yeah, we so that's, basically the general. You know the general rules we try to um yeah well. This is why maybe here's a good example.

A

Oh yeah, here it is, it's syntax contribution, testing.

A

Oh, what the hell happened here.

A

Oh god, oh no! What happened here all the api stuff somehow is listed on the top level. I hope that's not happening here.

A

No, okay! Well, that's good! We need to figure out what happened there. That's! Oh! Maybe this changed with the docs command. Somehow, that's really weird that shouldn't have changed at all, though that's weird, okay! Well now we know it's uh okay. So could you actually open an issue for this um sahil? um The the fact.

G

A

Yeah, it looks like the references are showing up. um Well, you know like they're, not nested anymore, for some reason odd. um I don't remember seeing anything that would change that.

D

A

D

A

Yeah, the the api, I would say, like you, know, docs um api or something is um at the wrong level. Yeah see it's at the bottom here. Okay now I can't mouse over, but you can see almost where my mouse is right, right below that you've seen apis, slash, master api service, so for some somehow everything under slash api got blown up into the level above it. Oh, I heard sphinx or doc. Utils got updated recently, but that shouldn't affect us.

A

I think it must have been that pr, because on the old branch, it's working, okay, we just need an issue to track fixing this.

B

Before release.

A

That's a big one: um okay, so be split out into its own plug-in um or its own uh package within.

A

Well, org: let's.

A

Not use um things, let's not use the helper from there.

A

Yeah, the idea is for that to be sort of its own project. You know.

A

Let's not use diff mile helpers in console tests since console tests.

A

Since that would be, you know, this would be a test. Dependency of dffml um console test will be a test dependency, oh dear, well, so not available.

A

um Two main package.

A

All right um since what uh let's not use, two helpers and console tests, scott's test will be its own plugin um and that would create a circular dependency. um Okay, so we'll! Let's talk more about this.

A

Okay, let me just I want to get a few more things. um Okay, so let's, let's we talked a good amount about the stuff, so so we need to do some more review on this. This is basically we need to review this, so I'm going to grab everybody else's items and we're going to come back, but this is a great enumeration of the things we need. We can go back and see whether we need to talk more about any of those things. Okay, so um and then okay, so we got linting of keep here.

A

Common message is time, decorator for locking good idea.

A

And then we got 10.78.

A

Okay, so and then hashim so do we have. Then you have a pr up. Let's see.

A

So hashim, what do we have for you today.

G

uh Yeah I started working on uh the cashier download issue, uh but uh it's not yet uh up to the point to get it reviewed. Okay, um sweet great job! No, because I have exams clean.

A

Yeah yeah all right yay, oh my god! This is such a mess like. How did I do this and not see that this was just the most insane thing in the whole world I keep. I did spend so long on this and I'm like wow. This is nuts. Why am I doing this? Well.

B

A

At least it's in the git history for anybody who actually needs to write some kind of insane um insane, um uh what's called decorator, that actually needs to take the function. um Okay, let's see yeah. How did I not realize that was nuts when I was doing that? um Okay, great, let's see so all right? So we'll just say this is a work in progress.

G

uh I I wanted to ask one thing: uh is it okay uh that if the both of the functions are just uh you know, returning the path to the file that has been downloaded or unpacked.

A

Yeah, I think, let's see, let's see, I believe so. um I think we're also gonna want to make them. um I think we're going to want to make them async actually we'll just put so so I I realized there were some comments about. um You know it's. It's sync anyways it's written somewhere, but um it's synchronous right now and but you know, a network operation really is. Is it's very much less than ideal for that to be synchronous right? So, let's tag them with async def.

A

You know instead of just def and make sure that they're awaited everywhere and that way um that way, if we actually ever do change that you know. Hopefully we can have another issue, then let's just put this down so cash download decorator to function pr, oh, actually, that whole thing was not a waste of time because of the operations could use that massive um thing. Okay, so cash download decorated to function pr all right.

A

A

All right um and let's say okay, so uh let's make uh cached.

A

The cached download functions, async def.

A

So in the future we can open an issue issue for in the future, converting them or like running them enough threat or something for the future um running downloads.

A

In a thread or something to make them not block the event loop, all right.

G

Yeah, actually I saw uh it awaiting that extractor today and uh I thought that makes sense. So I already started doing the async thing.

A

Okay, great yeah, I just I saw I saw that yeah I saw, or I saw your weight extract and I was wondering I was that's what made me think of that. Actually, so that's funny um yeah, okay, great great we're on the same page, then all right, um let's see so okay so and then and then you had um so moving moving models between notebook pr um do you want to talk anymore about that? Is there anything to say on that today or I know your your proposal has been sort of.

G

A

Of flushing that out right so.

G

Yeah, I'm just gonna, you know: do this one. First, the cashier download get the test running on the right. Notebooks.

C

G

Yes, that's right.

A

That's what happened here great okay, so I'm waiting on cached download to make testing of notebooks easier, all right, yeah- and this is you know- and this is sort of did you have anything else you want to talk about today.

G

No thank you all.

A

Right, great- um and so this is kind of um you know this is this- is this is sort of the main reason why we have these cache download and all that stuff in net is because anytime, we need to mock stuff. You know, um we know we have a defined set of things that we're mocking um instead of like doing sync, you know doing url opens everywhere. We we always go through these helper functions and that helps us uh audit and then do right tests in a more consistent way.

A

um Okay, so general business, that's hilarious, um okay,.

A

Okay, this is work in progress. This one is waiting. um Actually, these guys, I don't think, are quite done. Okay, and this is not done.

A

Okay, um let's see who else do we have today right? We have sudanshi, sir tanja. I think we missed you last week. um I think I think you didn't get a chance to talk. So, let's make sure we we hear what you have to what your update is. So what what do you have going on.

F

uh So uh I'm actually still working on the ice cream demo.

A

F

Saw that work in progress pr yep, so in that I am not able to actually uh like uh work on the merge command for it. ah I was not able to figure out like how to uh like run the merge command with a data flow source.

A

Okay, yeah: we need to document that okay, so we can do that right now. um So, let's see and then.

C

F

A

On okay, I was going to try to do a rebase of master against the accuracy branch um and and see where that goes. I've just been like I had to take a I had to take a few days off um last week and that's why I was m.I.a um but um yeah, so I haven't gotten to that yet. But but I want to do that um because that would be great. You know, I think we're almost done with that right.

F

A

F

A

Okay, so let's see accuracy, uh pr is to have master rebased and okay, and- and I was I'll look at that- so don't worry- don't worry about this until I just wanted to put it here, because that's related to your work.

A

A

Okay and then so need to figure out, merge or need to document merge command, and so so did you get? Did you were you? Did you make an issue for tracking the docs thing that we found.

C

I I just made her.

A

Great thanks uh what what's issue number.

D

A

All right great thank.

C

A

Awesome, okay docs seem to have exploded api into top level. Okay, great, we'll just try to fix that. It's not critical! We fix it now, just as long as we do it before release um okay, so ice cream demo, we may just do a release before um actually we'll probably do a release before I'm moving, I'm moving groups- I think I said so. I still have all my uh um all. My compliance stuff is within this org. So I'll probably try to do a release this week.

A

um Okay, so all right, let's see who else we have on the call? um Let's see uh natash.

A

C

A

C

My latest beer, that is.

A

C

A

Oh- and we already also started on the other ones- okay, great, let's see how those are going really inline strong starting strength. Oh my gosh, okay, they updated stuff with, uh I think they updated. um They updated docutils, and this is throwing up issue in another project, and this is annoying a lot of me.

A

um So inline strong, certain string, let's see 5060 or 85 260, was it.

A

It's what's this, it uh and my strengths are 5260.

A

Okay- oh that's because.

A

We importance type okay, um let's just fix that then, okay, so.

A

Actually, okay, let's see ah man, I hope that's not gonna, be.

A

Why do I keep getting the tests.

A

Okay, so what yeah? So so, if you guys, are seeing random stuff with with doc utils or like with the docs built, I I think that the the ci jobs may be. um Where was that um the ci jobs might be a little finicky?

A

Well, I'm not seeing that at all wait a minute. Where was that coming from.

A

Oh, this is above lgbt. What is this? This is something that exists oh great here now. This is lgbt.

A

Classifier um classifier.

A

Args, oh because we're doing yeah we're doing the make make class okay. Maybe we need some way of escaping things like this, because yeah see.

A

This making them by config, so this is this- is probably going to be an issue for more than just this um because make numpy config um just spits this stuff out, right or- or maybe we probably probably need to do this under. um We probably just need to change the way that where were um we're doing these guys. Actually, this is something that I wanted to talk about a long time ago, but but okay, so so the way that we do these these doc strings and stuff.

A

um This is a good issue that that somebody could could grab, but so we have these config structures right um and- and you know we we've have things like the make config numpy may config in spec to take from a um and let's just pull up the the test for making fig numpy, so that one is really obvious. What we're talking about.

A

A

A

Is wait, it's yeah all right, so this is. This is a numpy style, doc string right that we're parsing out and for the for the lgbm classifier. um It basically parses this and turns it into a config structure. And that way we can. You know, access this stuff in like a programmatic way, um so because it exists as as a data class rate. um So.

A

This the what we'd be doing here is, is we may? Actually? I guess this is this- provides two options to us. So now, okay, we have. We have two options right, so so we've been defining these things as config structures, and now we have the ability to take doc strings and turn them into config structures. So we could write. We have two options here. We can write just the doc string or we can write so we can write the doc string and then use make numpy config on you know the object.

A

So if we were to do oh, no, we don't want to do that. If we were to do um you know if we, if we want to call make numpy config on this double ret right- and this was our our model right. Imagine this is a model class.

A

Then um then we could write just the docs string like we would. You would usually write a doc string in a lot of projects. You would write it like this right and then you can call make numpy config on it, and then we would have our dfml style config, structure right and that allows us to have programmatic access to the whole thing now. The downside to this approach is: if we did this for every model. The downside to this approach is now.

A

You can no longer do this thing like we did here right and we do in a lot of models where we're wrapping the model right and so we're we're, taking we're taking the the lgbm model and we're reading its dock string and we're creating a config structure out of that and then we're adding more properties to it right.

A

So if we were to just if we were to write, if we were to to to just rewrite the entire, um you know if we were to take the doc string of the lgbm classifier right or the lgbm model um from from the from from lgbm the library itself that we're wrapping right now and we were to put it in um and and we were to just paste, the docs string in here, we we would have to keep that up to date.

A

First of all, and if we programmatically grab the docs string and put it in here then well that doesn't really make any sense right, because now we're now now now we have a different. Now we have a different um set of docs strings, um because this is, you know, that's not applicable to our project right. That's the that's!

A

The code, we're wrapping, so that's not quite the same usage and everything um so the the the alternate the alternate solution here is is to do something where we could like rebuild a numpy style, doc string um from a config structure, and now this allows us. So basically you take this numpy right. You created a config using a numpy style docstring, you added properties to it, and now you have this programmatic.

A

You know, structure that you can use and you can serialize and all that good stuff right and now, if you wanted to- and the point of this is- and this may not even fix this, but we can at least do proper sanitization.

A

So the point of this is the way that we've been doing this right on on the on the plugins page. um We have, like you, know a different. It's it's not the same style right. uh So if it's not the same as like the api generated docs.

A

um So if we were to go future feature, okay, that's obviously a bad one, all right yeah, I'm not sure if we really do that many places, um but if we, if you were to look at where record no, I don't think we really do this anywhere. No, maybe the net stuff util net yeah parameters right so it'll format. It it'll format, things like this. If we were to write it, um you know if we were to write our docs doc strings and numpy style docstring for the classes that we've been doing.

A

It would format it like this.

A

So what I'm suggesting here is that we could have a we could have a function that takes a config structure and turns it back into an empi style dock string, and that way we end up seeing you know this under the plugins page, where right now, you see you know the custom formatted situation, which would be. You know, and this is a bad example. No, it's not all right which is like this right, so you end up seeing it in in the standard format.

A

Instead, um this is an option, and the reason why I'm talking about it is because this right here is coming directly from the dock string of that lgbm model, and so we could then escape this like it's been escaped here for some reason, but it wasn't escaped here um or we could just modify the um or we could just modify the the the docks, the scripts docs api uh or like doc. I think it's stocks, py yeah, doc's py, um to escape it there. uh This might be the better approach.

A

This is probably the better approach immediately, um however, sort of long term, it's probably better. If we go to go to this thing, where we would have this function, that would be you know, like the reverse of this, make numpy config um anyways, that's probably an issue to be created. um I don't know and and also sort of something that that I think is worth asking. You know the group here.

A

Do people think that it is a good idea to um do people think it's a good idea to change from this format into you know the other format which is like this right.

A

Does anybody? Does anybody think that it's better the way it is right now.

F

B

F

A

F

A

It should be like this is what everybody thinks.

F

A

All right, all right, okay, I completely agree um so, which was the point of that whole rant. um We just obviously you know having two things. It means that we end up with two two approaches there so and then this should allow us to do something like um this should allow us to. um When we're um uh we we can. We can basically yeah we could we can. We can write this doc string, you know and then do you know, um make numpy.

A

You know from config and then pass the config structure right, um which would be like this right and then this becomes an f string um or something like this right um and and and this way now, we now will end up with stuff that looks correct right and we'll have the right config structure. The problem is, you know the the the the unfortunate thing is that we end up in this sort of in between land right where we have to do both things, because we need the ability to.

A

We need the ability to parse out the parameters from anything that we're wrapping because we're in this world, where we're wrapping stuff and we need the ability to add our own parameters, and so we we have to have some way of keeping that in sync.

A

So if anybody has a better idea on how we might keep that in sync, um please you know please. If somebody could create that issue, and then we could comment on I'll put the issue in once somebody's created. Who doesn't? Can anybody or let's see I'll, just create the issue and then and then I'll have to um because then we can sort of- and this is something that we probably need to talk about long term right.

A

um So if anybody has other ideas on how to do this, then util config function to make um numpy style uh arguments. ah No, what? Oh god my computer might reboot in the middle of this. For some reason this is it started this like three days ago.

A

I are you serious right now and then it was like you're gonna need to reboot. Well, okay, I meant lose you guys, um but I'll I'll come back. Then. Okay, so function to make numpy style, dog string from config class, and so if anybody has any um see recording from 2021 0, 4 13.

A

um Okay, so if anybody has any ideas on how to do this in a better way, uh please comment an issue because, obviously this is that is my entire thinking on it, and if anybody has any other thinking on it, does anybody ever have any other thoughts on it right now,.

A

Okay, maybe two: this is a good one, maybe for us to target for beta.

C

A

Oops all right yeah, so if anybody has any and I'll label with the community, input needed all right and I'll pin it, um because this is something we need to figure out um and I don't think this should be too hard. Quite honestly, I think this sounds pretty dang easy. um We just. We just need to do it um for this immediately and why we were talking about this is that uh I think to get this merged a test we're going to need to.

A

I wonder if I could just do it now, but sanitize default now. This is typing information, so this is unfortunately, what has to be done to get a config structure to look like restructured text, um and it's not pretty um so where is the uh because it would be way better right if we could just have standard standard stuff um it's in here somewhere. So basically, this is the like. You know, where's the built we built args um help plug and help um replace.

A

um Star star, we can try this.

A

We can try that and see if this docs build passes. Will we talk about something else all right, so it could be our quick hack, fisks, so lgbm.

A

I really hope this thing doesn't reboot me in the middle of our meeting. I cannot believe it did that jesus okay.

A

And then the other thing that we're looking at here is this uh async test case stuff, so consolidate test case classes.

A

All right, um okay, so and we're sort of let's these are these are sort of going in parallel, so um natasha is there anything that you wanted to so so we're we'll de-bug these okay make out oops.

A

So we've been debugging these things in parallel, the entire or because concurrently, the entire time we've been talking, but.

B

um Anything else you want to talk about today.

A

You're on mute.

A

Oh, no! No! No! No! Okay! Well!.

B

I'll see you guys soon.

A

um Or you know oh gosh, all right. Well, this sucks um all right. I I'll I'll, just ping you on gator when we get back to this, because I think maybe we should do something.