Node.js Node-API Team, 15 Apr 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Node.js N-API Team meeting - April 15 2019

Description

https://github.com/nodejs/bootstrap/issues/114

A

So welcome to the node.js community napi team meeting before April 15 2019 we're gonna, follow our standard sort of agenda, which is to ask for any announcements and then look at what we've got tagged for milestone 11, so the kickoff. Does anybody have any announcements they'd like to share.

B

A

Okay, so let's take a look at what we've got tagged for milestone: 11 I'm, just going to go there. Okay,.

A

So the first one is no js' blog post I think that might actually be complete right. Yep it's live, so that was a good blog post that went out and one thought I had is you know the people were talking about. What we should highlight for 12:00 did a number of the changes related to that is 12. The first thing they're shipping in um you know actually, okay,.

B

A

There any chance you could like send me the you know if there's a couple PRS that went into what 12, because then I'll say: hey here's a few things that went in and really it's all in support of this effort, which is one of the things we should sort of highlight that progress was made on well sure.

B

I can look at I, can look at 12-point, X staging and see see. What's in there, okay,.

A

Yeah, if you just caught, you know that'll, let me easily say hey. This is the case for why this should be one of the highlights. I think you know it's a good way to highlight any API as well again to say look for make we've been working on making progress across releases to make it easier for know that ons, you know, we've been working on API, of course, there's improvements in any its version.

A

1.4 is part of that and there's these other things which are which have also gone in which you'll send me links for and hey here's a great blog. You can go read more about it. Sound good.

C

Okay, okay, so otherwise, though,.

A

Will close this issue right.

D

The screen- yes thanks, sorry forgot.

A

A

No way you don't have to just look at my face: okay, so I closed, the blood hasn't the blog post submissions. To note coffee, you.

A

Think we talked a little bit about that I.

A

Did talk to Matteo a little bit, it sounds like yeah there. There might be a little reluctance to to do a similar workshop, so the discussion we did have about you know framing it a little bit differently. I think is important here, I'll flip over to that one too.

A

Actually, sorry, oh, let's close off for the first one, but um anything else. You think we should talk about in terms of the submissions to note comfort.you itself, I.

A

Think we've got a couple there right, so Gabriel you've got one which is lifetime to native add-on. That'll give us some coverage yeah and then Nick was gonna. Talk about one you know using any tie with different programming. Languages. I think would be quite interesting too.

A

And I guess I think we'd leave it at that. The last I can't remember last year, I think I think there was still Aruna should done a talk on an API I, don't know who attached, if there's still an interest to do that or I.

E

Don't think I'm not just planning a talk this time around, but I can check with him I guess.

A

It's really like do we all feel comfortable not like this would have a there be a couple talks which would mention any PRI potentially, but wouldn't be the focus of hey. What's going on with the API effort, I.

E

Think that's fine I think, like the project is at a different stage now I think last year it was to bring awareness to the effort and I feel like there's awareness. Now and now it's just hear different ways. You can use it and engage right.

A

I think we still need to get it, but yeah the try and get that indirectly, as our best bets yeah.

E

A

Okay, so I think in that case you know those two like very good candidates on my mind for that and I'll. Take that one we'll leave it a milestone. Okay, so the next one was the workshop right.

A

So Jim put together- and he did say he wasn't gonna be here this week, but he put together this abstract.

A

There was a few updates I.

A

Think it looked very good to me. I just had one small suggestion about like trying to make it so that people, even if they don't know anything, are happy to come.

A

I've also heard that the Jas Interactive is going to open the call for for a submission soon, as well.

B

Did we ever established a venue I montreal for a jeaious, interactive yep? Where is it.

A

Montreal it's in Canada ôhe.

B

Montreal, okay, I might.

A

Say all right, I was gonna, say what it's like. You came from chronic. Let's know how much real is no.

B

B

Bought you all this, but I didn't know. Jas was moving there. I.

A

B

A

I think I said you just didn't hear me when I said it's in Montreal.

A

B

Okay, all right yeah.

A

B

I got it okay, so so suggest interactive is in Montreal, that's cool, yeah central is Canada. It's.

A

About well, actually, there should be direct flight which flights from Europe, which is good like for me. I have to go indirect, but indirect means through Toronto or Montreal, so I'm pretty sure there'll be direct flights to a lot of the major European hubs anyway, yeah and probably about six.

B

To seven hours.

A

B

It's a little further east too, so so that's good for Europe! Well,.

A

Yeah exactly like it's definitely four hours closer than Coover, three or four um so I think we should submit this one as well. I! Think I guess we should just wait for Jim to come back and then you know, unless anybody else has other suggestions, we should just actually submit this. One sounds good along with the comments of hey. You know this was really really helped us last year, so we're really hoping to get it in again this year and I guess we should also I wonder if there's like anywhere else we should be proposing.

A

This is a workshop, because I think that would be you our biggest bang, for the buck as if we could get this workshop into say. Other conferences.

A

So they're right now.

B

A

B

What what about guys is there gonna be a note summit cuz it's dead soon, like a good place, for it definitely.

A

We would do that. We're still waiting to hear I haven't heard if that's happening or not I'm, starting to wonder yeah.

B

Yeah yeah it's a little late in the year.

C

I saw these news conference and the icing on Twitter.

D

Yeah, so that's in February next year, so that's I, think possibility.

A

A

Guess some of the Jay s cops are probably a bit more front-end oriented and this is pretty pretty see. Seebeck ND, ish, yeah.

B

Yeah, that's true: yeah we'd probably be out of place, but I, don't know, I mean um I, don't know for for things like for things like SSR might sort of tail.

C

B

This I don't know, I mean it certainly can't hurt to try right, yeah.

A

So yeah I think if everybody can kind of think of conferences that might make sense and propose, is that I think getting into one or two more like in a workshop type environment would definitely be good for us and for the effort yeah, because we got a pretty good bump from the note coffee, you and I think something similar a few few more times somewhere else. I think might would be good.

A

A

Wonder like cuz yeah I'll reach out to Joe cuz I know in New York. There was something in like the September timeframe last year.

A

Okay, oh yeah. Well, we'll keep this one open until we get submitted and also people can. You know think between now and next week to see if there's any other options.

A

Separate additional package on top of no dad on API the test you put together a draft of what that would look like you want to take us through that or.

A

Or, what's what's the what's useful discussion should we have this week.

E

Yeah I think the last time we talked about this. The plan was that y'all going to review it. So I guess I, don't know if people have had a chance to review it and looked at our have specific questions.

A

I've read through it, don't have any specific questions, it's kind of like how do we make the decision right.

E

So the suggestion I had was to engage with no data. Api users- maybe open an issue there and ask for feedback around which options people preferred right.

E

But I don't know if somebody else has a better idea than that. I.

A

Think before I did that I want us to convince ourselves we're happy with either option but I like the real you know, I could see as a consumer if it's all in one there's not really much of a downside other than space I guess that space downside then the biggest concern in my mind is the is the the expectation that this group would maintain. Whatever goes into the core one right, yep, and so it's I think we need to say yeah if you know we're comfortable with doing that and ice yeah.

C

A

Really, where I think this, otherwise you want to go and ask get asked a question. You get the answer, then then say wow. We're just can't do. That is.

E

This something that's worth getting PSE and put on.

A

Things I'm not sure we're gonna get a lot of feedback because the module is you know it's if it was in the core or API is itself I think we get feedback, it doesn't hurt to ask I'm, just not sure we're gonna get an answer that Cliff clarifies it for us, like the TSC model is really the people doing the work, we'll figure out the answers right and if there's a, if there's, if there's, um if there's uh you know, conflict or whatever they'll help resolve it but like in this case, I think it would be like hey if we said we're perfectly comfortable that we can handle all the work and likes bring stuff in then they've probably score for that.

A

If we were saying well, no, we don't think we can. There also be support for that right. Yeah.

B

Yeah I think I think I think the fact those things are heading towards like the the ecosystem solution, because because there are, there are like there's an issue where, where people are mentioning like I, can that be threat, safe function, implementation and stuff, like that, like a C++ wrapper that that is out there somewhere.

B

So so people are writing these things and they're not waiting for us sort of you know so, and that's good right like yeah, it's just it just sounds like you know the people, you know, vote vote, vote with their feet kind of thing you know and they voted for an ecosystem solution.

B

You know they just write these things and they throw them on NPM, because it's so simple right so I'm sure I mean we could we could start to like curate, maybe, but but aside from that, I I, don't know if you know I mean I, don't know if we should. You know make an explicit effort, because we might end up duplicating some stuff. That's out there, even potentially.

A

I, like the curating side of things like her, providing a place where people can say hey here are these things that we put together right, yeah.

B

Yeah, that's that yeah, so so, basically, just just sort of you know keep keep a keep a list and try to keep it up-to-date of things that are out there, that that also wrap an API or also add C++ based functionality. On top of any API yeah, that's yeah.

A

That's not like you're here you.

B

Know PR yourself in yeah, basically.

A

Yeah, it's and then we'd be aware of it, which is good and we can point people. You know it's not that we endorse them, but like here's, the list yeah. If we see that there's one- that's like. Oh, that just has to be part of core or then we can. You know pull it in yeah yeah.

B

Okay, so so yeah, if we, if we, if we mentioned stuff like on the on the read me page for know that on API, like additional packages or something thing that yeah that's eventually a way to go and.

A

Then so, if it's something which is I mean so then it's generally, if somebody comes with something which is beyond a wrapper of core, the first suggestion will be, you know, create a create, a module, and you know add add your references here, you know, will help publicize it but create a separate module, and then you know on you know things like the where that's not possible or you know. We really think it needs to be in core more the special case, like TFS n, where it's like.

A

Well: okay, that's not necessarily that's still fairly yeah. That is that is this for okay. So we got that and then things that we really think at some point we could revisit. But to start with, it's like a start.

E

That's not like a fine approach to me. I think. One thing we should be precise about is then our plan for API support and like deprecation and so on.

A

Well, yeah I mean I, guess we we do yeah, we haven't said we won't change the API. We haven't actually made a breaking change yet.

A

But yes, okay, so you say basically we'll give this amount of warning or whatever right.

E

Right exactly I think that should be good enough in terms of an ecosystem. If the ecosystem pushes back, we can review yeah.

A

And we really try not to in the first place right, like we've gone out of our way, not to actually change anything when we didn't have to yep.

A

Okay, that does that sound reasonable, like a reasonable approach to everybody.

C

Yeah I think that the ecosystem approach is more flexible than others, so yeah I agree.

A

Okay and then you know if there, if we come up with like some in practice, problems or you know it's like some real frictions being introduced, because that's the way we're going then we can think about. You know other alternatives.

A

Now the two pr's above I guess the thing we said we got the first cut, taking the two approaches, the next ones, to get input from the two submitted, those to PR. So maybe we should reach out to those two and just say this is what we've decided right.

A

So maybe I'm going to say documents.

A

A

Document of preferred approach, so, basically I think the next step is for us to write down either in this issue or I. Think a PR to know that on API would probably make sense.

A

Anybody want to volunteer to do that. Do.

E

You want me just update the doc I have and link to it in both those PRS yeah.

A

Although I think I think like as opposed to having in a Google Doc, we probably want the text in the repo itself linked off the readme. That says you know we considered these two approaches. This is our you know. Maybe it starts out with like here's the approach. If it's something that wraps it will take, you know we'll take that into core itself.

A

If it's not wrapping stuff, our general approach is to ask that you do it in an ecosystem module here the two, the two options that we considered, which would be pretty much the text that you have and in and then probably up front as well. You know here's a link where you can PR in your module so that it's visible and other people can can find it easily off the real.

E

Yeah I can take a stab at that. Okay.

A

That sounds good cuz. Then you know then we'll end up with like actual documentation and the repo that we can point people to say to say: here's, here's what it was and here's a rationale right, yep.

E

A

Can probably also have a I, don't know? Is there any more real there's, not a lot more discussion in the issue itself that we could link to so I? Don't think, that's necessarily worth yeah! You can take a cut at that. I'd be good.

A

Okay and then I think we should send it to the people who set a couple PRS just saying you know, as on the PR, we should mention them at mention them saying hey. This is this is what we refer to approach and give them a chance to comment on that and then go from. There sounds.

E

Good okay, good.

A

So async worker, so yeah I, think we can spend whatever time we have left looking at the next part of async worker.

A

Are there any I'm? Looking at the like issues burned down the list of issues in the repo the one I might call out is.

F

A

Sorry I'm trying to find the one that I.

F

Don't think that was it.

A

But my email, cuz I, just want to highlight it. Cuz I did somebody did that the originator did basically reach out to me, saying: hey still haven't been able to make any progress on this one.

A

C

Yeah the Donora talking about you right.

A

C

A

Were they reached out to me and said: hey, you know, I can't can't make any progress and actually we're even willing to pay somebody. If you know somebody who would help out so I thought I'd at least bring it up too, so that everybody knows that they're they're definitely eager to get some help.

A

um But at that I think that's about all I can I can say at this point, because I really haven't had a chance to do I. Don't.

C

Know Nick, if you have any.

A

More insight, what's going on there or.

C

C

Because he cannot expose.

C

C

Since seeing all seems good but I don't know, what's happening.

C

C

I don't know he.

A

Okay, so the reek so the Cody half there doesn't recreate the problem itself. Yeah.

A

Yeah that was make this makes it.

C

C

A

We really needed to recreate the code that recreates and then we could look at if it's a Windows thing attached might be able to help out, but.

C

I think it's not maybe it's.

A

Right, okay, but really I, guess it's!

A

You know it's. Basically, we need to get.

A

Either I guess it's somebody like I guess: I'll respond to saying like either. We basically need some code that can be shared publicly, that recreates the problem or like I might be able to suggest. Like you know, one of the consulting companies like I, don't know if near form does would do something like that, where they would work with them and they could sign an NDA.

C

B

Yeah, we should also be careful not to recommend anyone particular or sure right, yeah.

A

I'm, just thinking of any ones that I know of yeah yeah, like.

B

A

It's whoever we could open an issue, yeah, basically saying somebody. Who's, got some expertise in this area or a node that would be able to help out.

B

Well, we could we could, we could serve dish, ously mention hotel or something right. There's.

A

Actually, a few people that you know I'm thinking of like individual, who could be like individual contractors, might be interested yeah.

B

That's another good point: yeah.

A

Oh it's kind of like maybe we can make a post in members and basically say here's an opportunity for some paid help yeah yeah most fair way to do it. Yeah.

B

Yeah, that's right, yeah, because we just can't see it otherwise. Yeah.

A

Okay, yeah I'll, so I'll respond to him with that that basically like either we need, like you, know, a recreate that we can see publicly otherwise, like the the other approach, is that you know with you, if you I suppose, if you own a no track, somebody might help you and, or you know, we'll post an issue in the the members we could post an issue in the members.

A

Basically saying hey, you know looking for some help on this run, willing to pay hmm cuz, I guess I should mention like you know, IBM sells note support and that I was just thinking like you know it's to take money for a one-off, but if you actually bought a support contract from somebody, then they'd state helped you, okay, okay, any more anything else to discuss on that front.

A

No okay, let's see that's the main one I want to highlight before we switched over to async workers side of things anything else. People want to highlight.

A

I'll take that as a no. So let's flip back to that, because I know we had a good discussion last time, let's see where.

F

We can continue on from so.

A

Okay, so last time we got to the point of discussing the unsafe destructing behaviors reference counting and I think we came up with a good solution on that front. So, let's loop back to the concerns which were.

A

Okay and it's a cancellation, so I think we've covered that first one. So the second one is hard-coded: gems got JavaScript function, callbacks.

A

So is there any reason we can't just make that an optional thing and if it's not there call it not call it.

B

Yeah that that does kind of look like an eyesore a little bit, because the it really should be the optional there's I mean the what you do yeah well. What what you do in the on okay and on error should be completely up to you. I guess, that's that's a little, but I can see it's handy to be able to say you can have these functions week.

B

All right, yeah, yeah, no I mean, but basically we're saying that we're basically encoding in this requirement, the assumption that you will call back into JavaScript and that's not necessarily.

A

A good assumption.

B

Right so yeah I mean we could make this optional and then just go from there.

B

Let me just see what it takes: yeah.

F

Let's take yeah so like yeah, you'll, probably DP to it, but I'll pop the code too.

A

B

All the constructors take this function right.

A

Not not searching at.

F

All on the proper way, how about like saying you stinker.

A

Right so this callback right yeah.

B

There's there's there's no, no reason them specifically for it.

B

Cuz async work doesn't have the concept of a callback. It's not like a thread-safe function, I.

A

Mean it's handy for the default like okay, when you're on okay is up and you this callback gets invoked or or it's like you don't have to write as much code. Cuz on error are on. Okay, it's probably comes from what was in man as well. I would assume, uh let's see.

B

Oh boy, where you at yeah.

B

Yeah, it also takes a call back and it's not novel yeah, it's true so.

A

I mean I, don't think we want to remove. You know we don't want to break the API anyway, but even if we, if we were you know, we stay consistent with man, we wouldn't want to get rid of it, but making it optionally nillable doesn't seem to be a bad thing. Yeah.

B

Except it's a reference right, so the idea is that it's not like we're not seeing in man. The callback is being passed as a pointer, not not a reference. Okay. So so that can be null right. I see, we've passed.

A

It differently enough that yeah.

B

Yeah yeah that that we couldn't really personal there I mean we could pass. We could pass like a like an empty function like a like an empty value where, where the end is null and the nappy value that it's wrapping is not so so you know we could. We could pass like nappy colon colon function, but I'm not sure if you can give a default value to a part to uh to a reference. I don't know if C++ to know that.

A

B

I just make a second.

A

An additional set of API is that yeah.

E

Should not create like a separate async worker class, which does that you mean, like a you, mean like another constructor that doesn't have that yeah or you could even just duplicate the class or sub class or to some degree async worker. No callback.

A

Right, although I, wouldn't, why make a separate class versus just you've got an act? You've got additional constructors. You could use.

B

Yeah I think the thing is, though, if we, if we do additional constructors, that we'd have to do the whole set again, but that's okay, like okay,.

A

So, with the overridden one right, okay, cuz right, you'd have to give yes we'd have to give the whole set yeah.

B

And and and I, don't think we could yeah. We couldn't reuse the constructor that we have because like where would you get in the reference? Well yeah, I? Guess you could have like a dummy function? What why.

E

Not you could have a dummy function as a static on the class itself, which is you could also make that as the option you pass in the dummy function. If you don't want to provide a callback.

B

Yeah that would yeah so so so yeah just have like a nappy function, that is, that is constructed with no parameters and I. Think we and think function has one of those has a constructor that has no parameters and yeah and then just pass that one around we'd have to basically make.

A

Sure that that then was like we check against that, and we wouldn't call it yeah.

B

Exactly yeah I mean I, don't know where: where are we actually? Where are we actually calling those like we're not actually using the callback anywhere I? Think.

A

On in on okay, no.

B

Like on okay, we're calling we're calling a native callback right so yeah, so oh yeah, no okay, callback, all right! I! Think! That's that's my guess!.

A

B

Okay, so so, basically, basically by default by default, we we make on like an empty function, call like to JavaScript, it accepts no or it passes. No whole parameters into JavaScript. It's just basically like call back bracket back brackets is what it looks like on the JavaScript side and and if there's an error, we call that javascript function. Okay, so with the error right.

A

Yeah, basically, yes, you know you, you can provide a JavaScript function and you'll either get a you know. You can tell based on the whether it's yeah.

B

Yeah yeah, that's okay! So that's our default behavior! So so, if we added, if we added the dummy, if we added this dummy function, then then then the on okay Andy on error would basically just check if the dummy function was empty and and and all only call into JavaScript. If it's a real JavaScript function, that's all so we just wrap these. We just wrap these two calls in an if and that would be it exactly and yeah I guess we.

A

Could even provide I, don't know if it's C++ ish, but we could define like a no function or no no call back like a define which then resolved into the static from that class.

B

Yeah, but that, let me that's the same thing well, I mean we don't want a real function at any point right. So so so you know this. This underscore called callback may be empty at that point like we're. Basically we're basically not checking, except for not we're, not know, checking callback itself, we're not tracking the contents of callback, because, ultimately, ultimately, even though, even though callback is is, is passed by reference into the constructor and all that it's still not a ball, because, oh because at its heart it's just a pointer right. It's just.

C

B

And if that nappy value is null- and it can be now because it's a pointer, then we can check to see that it's not all here and and only and only call one of its members, if it's not so so, essentially, even though it's a even though it's it's like a C++ reference, it can still be null because of the nature of the class. Oh, it would be like a function.

A

So it would be a right so like there would be a static which was defined as a function. That's on the class itself right and we don't know the contents like it doesn't like. We could even check that address if we wanted to.

B

Well, it would be, it would be a yes, it would be a static function, but it would be constructed using like an ol nappy value and a null environment right right now, because I guess.

A

B

Constructor and value for for a phone.

A

This is that doing a copy like that ad signed, isn't doing a copy right.

B

It's taking its passing done. Well, that's right! Yeah, yeah, yeah yeah, but but but that's okay right because I mean uh like like a static static. Member never goes out of scope right so right, but but I'm just saying we could actually just do. Did.

A

You pass us a pointer to that thing: yeah exactly yeah.

B

A

At callback equals equals this other thing can't we do something like that. No.

B

No, it's not you're, not checking the the value of underscore callback right because that's a reference right so thatthat, so that's never gonna be no, but but you can check like underscore callback, dot, dot value, basically, which.

G

B

The nappies of value inside it.

A

Okay, but I was just thinking you might even do a simpler one like it's always gonna, be the same address for that.

B

Empty thing, but but we're not we're not dealing with addresses, though right, because we're not dealing with pointers so so the instance is actually you know it's a reference so so.

A

Right, but is it a copy of the reference or if we take the address of it, do we actually get the address.

G

Even if it were to be a copy, you would have a copy of something that has a null pointer for the environment and the value yeah.

A

G

That's why yeah so, if you're actually making the copy, then.

A

G

You would see if the environment is no point, is null or something yeah.

B

And I think I. Think man that, like nappy value or I, mean that B colon colon value has like an empty method. So so we would just say if, if not underscore back is empty call back call and.

G

B

G

Could you just then have another constructor that does not have this function, but then you just use? Yes, you have to use this defined thing that you mentioning earlier yeah.

B

Exactly so so we do so so we create like, like like a static like a static member to this class. It's of type nappy function and and is initialized to nothing and so is empty, will return true for it and then, and then we use that we pass that into into the constructor into the existing constructors from the new constructors and that's yep.

B

That sounds good, and then we will have made the function optional and it's cember minor. At that point, right: okay,.

G

Are there ever any concerns about constructing these static objects? Now like, because now we're gonna be providing code that does create static objects that may or may not be needed? Oh.

B

Well, yeah yeah: it is kind of an artifact yeah, that's true, but we're already we're already past the point where we are we are. We are truly purely in line because we already have like static static functions which are the the pointers to which we pass into like the the C API. We have to write Ram so so so that we are, we already have stuff that is like real, rather than you know this. This was this was part of the reason why we had to introduce the F visibility equals hidden for OS.

B

10 is because we do have symbols, and you know, including an API dot. H in your code, will result in symbols being added to to your your your object, file and, ultimately to your library. So so, yes, we have symbols to our name. Basically, and- and so you know that- that's why we ran into this. This visibility issue on OS 10, and this is gonna- be another symbol that we add, there's just no two ways about it: gotcha, okay and.

A

I guess what you've described was it being empty? Is the safest one anyway, because even if there was confusion, because there were several, they both be empty and it wouldn't really matter right.

A

Yeah, okay, so I, just typed that as the solution for this one does that look good.

B

Yep, that's right: okay,.

A

um So that one sounds good okay, so javascript function call back done. No support for promises.

B

Yeah this this async promise we have. uh We have a PR for racing, promise, I, think and and I'm about to comment to Philip. uh He he also, he also submitted an issue about this running. Note thread idea to just have like a one-hour thing where you you're just have to be checking along on your own thread and and by the way I want to run.

B

Something in the note read: that's it's kind of like a variation on the thread safe function where, where the, where the idea is not that you have a function, object which is now thread safe, but rather you have like a one-off. You have a reference to uh to uh to a JavaScript function like a persistent reference, and you know you cannot use the persistent reference on your own thread, and so you have this like running node thread thing where you passed.

B

The persistent reference to a callback which you know will be will run on the node thread and therefore using the persistent references is safe. So so he had that idea and I was about to comment to him that yes great run with it. We're happy to hear it, but please write a library, because we decided not to accumulate everything in all that on API. So something to that effect, I was about to comment and and I think this asing promise is is the same thing like it's: it's it uses.

B

It uses a sink, which is no the add-on API wrapper. It uses the promise, which is also I, know that on API wrapper, but putting them together is a feature above and beyond, just being a rapper, and so it is. It is right for or for it's for its own NPM package or for for depositing in another NPM package that we'll have that deals with these enhancements. So.

A

Basically, I just typed this one, which is no sperm based on earlier discussion. Suggestion would be to create an NPM modules. It goes beyond just wrapping any functions, yeah exactly yeah yeah, okay, that makes sense and I think we're out of time for this week, but we can loop back to the next ones and the list I think that's good. We're making some suggestion.

A

I think also there's the exploration you know the resolution was suggesting something new and it may be that the suggestion is too like okay, that's good, but maybe that belongs in a separate, separate package. Just because you know we need something that maintains nine compatibility in the first place right right, yeah, yeah,.

B

Unless, for some reason, you.

A

Can't build that, but if you can then okay, okay, but we'll plan next week to work through the next couple, any other last things people want to discuss before we close.

G

Up for today, so for those items that we discussed, those are actionable and can be put into the code yes responsible for for doing that is.

A

It anybody who has time it would be great if they opened an issue and started working on it. Otherwise, you know if we get to the end of the list, we'll find a volunteer. Okay, I feel.

G

A

G

Plan you're, okay or.

A

Even if you know, and sorry, even if you don't want to do, the implementation is the great, if you have time to open issues in the no data on aprv poke, saying here's the context. This is what we think is the right answer. If you have time, please work on it because yeah.

B

Yeah yeah yeah yeah, yeah yeah, that's yeah! That's that's a good approach to oh somebody.

A

Has time to do that piece, that's great! If somebody had then has time to go on and actually start working on the implementation, that's even better if I suppose of them have good value. That's right!.

G

Yeah, your third excellent okay.

A

Okay, any other things.

A

Not great and we'll talk to you all again next weekend through github between now and then all.

B

Right talk to you later bye thanks.