Node.js Core Technical Committee, 13 Apr 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2016-04-13 Node.js Foundation Core Technical Committee (CTC) Meeting

Description

No Issue

A

B

All right, we are live, welcome to the node GS foundation, CTC meeting for April 2016, the 13th of April.

B

We're going to do I, stand up for everyone, that's here, and then we'll get into a couple of issues that are on agenda for those who keep up with how we usually do these means. That was not an issue I created on the issue tracker this week. We have enough people here, so we're gonna do a meeting anyways I'm in the light that v6 is coming out soon, we'll probably a couple things to discuss so first on the stand up is shocker.

C

Hi I, don't have anything nice to say, I just come into some issues and food requests as usual ah so and some minor security things an ecosystem. Next one is gone.

B

Like Colin dr. here so next should be even.

D

Yeah, the only thing I'm really gonna come in a few issues, but I also schedule the next onboarding session Friday. So next is your eye.

B

Hello I've been doing some assorted documentation, sort of things and various other issue and Paul request related things, I'm also leading up to be six and I've also been working on our no dias and github bought, or one that at least a couple of our repositories use, and hopefully we're interacting with the bailed working group a bit so hopefully we'll be able to get it on running on actual build.

B

Hardware like the website is, and also the our plans, and we should be able to do it reasonably soon to actually get Jenkins to tell the blocks to grab more it more info from Jenkins or something like that. Anyways and actually post github commit see I statuses from Jenkins for us, so that's going to be really interesting.

B

Next is rich drop.

E

Hi, sorry, you need to find my mute button. um I've been reviving and closing old issues. I've been fixing flaky tests and I've been trying to figure out. What's going on with raspberry pi 2 devices and CI they I know, the build group has been doing since doing some work on fixing them lately, but something's up with their reduced network capacity, connecting the local host of all things, and so I've been mostly an effective on that. But it's been consuming fair amount of time. Next up is ally.

F

So I've been I was at the vm summit last week and a couple weeks, I've been working on the v8 5.2 PR and since the vm submit have spent a little bit of time. Looking at feasibility of doing dynamic, ffi in v8, and I think next is Michael. Dawson.

E

Did we lose Michael all.

G

Right I was on mute, Jeff. Sorry, I scheduled, ran recent benchmarking meeting I've been working on moving the PPC begin d machines over 2, mu, 2 14, so that we have a supported platform for doing regular releases. I investigated one issue where ra the test on AIX we're timing out and work around in place without attended to BM face-to-face.

G

Did some miscellaneous review landing commenting on issues I work with Gareth on a benchmarking presentation, that's covering some of the work, that's being done in the benchmarking work group for nodejs live tomorrow and just starting to schedule. The next API working group meeting to discuss some of the stuff that we agreed as the way forward in the vm face to face next is James.

H

um Yes, so got the version 6 release candidate, one else get the release candidate two out, probably this evening, working on some other items, just kind of clearing out some long-standing a diplomat to do list a number for questions going through doing issue issue triaging neca. I think a.

H

Gun never lets hope you aint funny, sir yeah.

I

3 make it here, Steven yeah, so working on a few build updates. I was at being summit also, so I have 688, which is actually should be on the agenda today and also working on disable break it earlier. We're.

J

I

Next is Trevor.

J

All negative note is the domain post-mortem PR with documentation. I, don't have a list in front of me.

I

No I think it a light. One thank.

B

You had lost so I did miss some of the agenda cuz some of its and pull requests and I did not see that but I'm like to do that. While some of us discuss the agenda so James, would you be willing to discuss playing for v6 status updates and all that on that.

H

Yeah, so as far as planning for v6, it's really straightforward anything that lamp by time again execute the the cut will be included. The the actual release, I'm currently looking at the 26 it coming out. That's a tuesday keeping a very close eye on vh version spies to see when it goes stable. It's supposed to go stable. I believe right on the 17. Thank you. Can I be right on in there, I'm not sure ali influence.

F

It went stable about doors ago, oh.

H

Nice, so if we can get the version 5 landed in master, then we can get our include them in release kind of this quickly as possible other than that pretty much Oh. What yeah.

F

So so I think I mean for the next RC I mean if it, if you can't get 500 landed before then I think would make sense to wait on the other see to make sure we get feedback on five hundo.

F

So we need ltd. Ms, on the on the purpose for 50.

H

So I mean yeah, that's that's really it I, don't think, there's really any any issues at this point other than the v8 blending up I know that there are a few similar majors that are still kind of pending to see if we're going to land in time.

H

At this point, though, I don't think there's anything that would hold us up if it isn't mine I think we're just to go.

B

So you plan on cutting ones you plan for the full release on the 26 on the 26, so that gives us a boat and not quite two weeks to test v8 4.5 kept me right. We go ipod. Five-Point, oh yeah, is everyone here comfortable with that.

B

Seems seems like an okay thing to me.

G

Yeah I wonder: do we make it's sort of the the first time we have like a 60 stable? It's a little bit different than a 50, stable right, I, don't I, don't know if I guess we make an announcement, so people know hey if you're you're taking the latest a little bit of a bigger jump.

D

Well, it's not me LTS yeah, right right.

H

Dad 60 that Hill will not be able to yet until next October yeah, the 60 will pull on the website. You should see version for still, as the LTS and version 6 is the latest stable. It's I, don't think it's really been decided how long we're going to keep maintaining version 5, but it's it's not LTS, alright! So right it.

G

Will really end up being from the download the front download page anymore right right.

H

G

Oh yeah I'm, not sure, is that it, you know, there's an issue here is just thinking through that. If somebody is, you know, consume stable regularly this one. This event is slightly different than any of the other stable releases, but it's probably obvious from the name. So maybe there's no issue what.

B

Event is is different about.

G

Well, when it's when it goes to six there's a lot more change than there's been in any of the previous tables right.

B

Well, that's because it's a major like.

G

B

6 is not on an existing stable branch. It's a new, stable branch. This is not a release on to be five. It shouldn't be considered create so that this is not like assembler minor upgrade that's on b5 and we need to message it as such, like this is a new, stable, yeah I think what we'll do is probably definitely do you know, maintenance related stuff to be five for about a month and give maybe an extra month for any big security things.

B

Something like that. I'll, probably what ends up happening, I think, depending on how much people interested in maintaining it I.

B

Think definitely at least a month and just a little bit for security patches, but I.

D

Thought we defined it in that document that we're all I dad I thought it was defined that it would stay. Can you get updates all right? Let me trunk on the docking I think.

B

It I think it's supposed to be maintained regularly for a month, I, don't think I'm, not sure I think it might get a little bit confusing. Personally, if we get like land lots of features on it, so maybe we can avoid landing all December minors on it, but still maintain it. Otherwise,.

G

Yeah I mean if you've only got a month to go, it's kind of like you need any of that stuff, probably not right. Yeah.

G

So the other thing James, so is it the 26 when will cut a new tag or a branch or whatever it's going to be.

H

Yeah so honest 26 for the new v6x branch will be created. Okay, alright and the version and master will be bumped to V 73 great.

G

Okay, I just wanted to plan for when I need to add a new entry for the benchmarking, graphs and stuff right.

B

um Would it be perhaps useful to already do that? Oh wait, you're already, testing master. Aren't you yeah.

G

Like there's already a master, so until we branch master is effectively the same. Oh.

B

G

Oh yeah I might do a little bit of work in advance to to get the numbers and whatever ready but yeah, just checking for when I should actually make sure. We've got something ready.

B

um We have any update on the I. Think we're done on v6 for now. Does anyone else have anything just caps on them.

K

B

B

Have an update on the legal stuff, or do we just need to shell that until next time, Michaels here.

H

Probably have to show that for Michaels here nuts, but there was another issue that I'm hoping one rod can actually be here. We can kind of talk through it, a bit where the board raised a question about clas and we need to have a discussion about about it. We're. Basically, the board is asking us to adopt a mechanism for having a CLA individual CLA in corporate ceiling.

H

We don't because the DCO, though yeah there's some complications with that that the DCO is copyright, only it's a fairly complicated conversation. So, let's, let's hold off until rod gets here, but it is something that needs to be on the agenda for.

B

Wait, who are we planning on having sign this, because we can't we're not having everyone sign that I'm going to oppose that and take it to the board in some way or other, like the it's, the most user on like contributor, unfriendly thing you can do is: have everyone have to sign something? We've.

H

Get at the board meeting the mid board meeting both lot and I raised that concern. Like I said, it's really complicated so issue. So, let's, let's not discuss it right now, yeah.

B

H

It's on the agenda for next time: okay, I'm, just yeah I mention it now only two to make it make ya'll aware that it is a discussion with some had noted.

E

Okay, at an open, an issue for please and let's move on yep I.

C

Want to say that our elders, right testis, license still have a close that and additional boxes and instructions from Penn website incorporates intro the license agreement, and this it looks like each hasn't been fixed on there in Pam's side in permissions. He has the British lion sense. When the latest information true is pleasant, an hour LTS branch, because the old license with the addition cross- and it looks like it- was not dissolved. Yet.

A

But I was this.

B

Resolved in four point four point: two, or did we not get an MPN release to it yet I know there was an MPN poor request recently, but it was a little bit late and miles decided to do the four point. Four point: three release without it I think.

B

Let me double check. I, don't think that mpm release was in, but I thought we already got their license. Ups in an upstream update from it I'm might be wrong. Now we.

G

Certainly has a relatively recent NPM, okay.

B

So four point four point: three has an upgrade from mpm and LTS to two point 15.1. I will link that issue in the chat and.

C

Bam vision, Joe still has the old license. Each hasn't been updated in the latest edition of the branch to.

G

So you think that it's updated in three, but not into.

C

Yes, I opened an issue for the xbox. It was our closed. I think that is mistake.

B

B

Oh yeah, I see um I thought, I was updated and I'm p.m. LTS will have to contact them. I on I.

C

Already do open tennis shoe at then in p.m. today, but it looks like it was called some way.

B

You think you can get me a length of the tissue.

C

You seen this Thanks.

B

um Who, let's, let's add, generally move on now because we have a couple things: okay, I think so the next issue is that may read it: where is it HTTP / server request socket, don't request that socket da bytesread is always 0 for incoming message on this is a bug and I tagged it for some reason. I think Fodor wanted to discuss it, but he's not here and neither is calling anyone else. um James did you look at that at all.

H

Which one other sorry look up.

B

It's the next issue in the meeting notes. It's let me copy the link that.

B

B

Boat, bytesread always being zero for incoming message.

H

Yeah, so this one, um basically because of the way HTTP parser works, it bypasses mechanism in readable, scream, readable that increments, by thread, so by thread, always ends up being zero on incoming message and there doesn't appear to be any simple way of fixing it. So the question is: do we need to drop by thread from incoming message or try to find some way of fixing? It am I. Think I'll you there that's the question.

G

This points red new or is it something that's always been there.

H

It's always been.

G

H

As part of the readable stream and it used to work on.

H

On the server on a secure server before, like I, don't know when exactly it stopped working, apparently but I from what the door. What I imply from what Fedora said? There were some changes to HP parser at some point that completely bypassed that whole bit of code within readable, so that when it's just not even sub par, riedl streams live in use, anymore, fluorine, chemistry,.

K

Well, I think the bytes red is actually at the net dot, socket layer, I, think and I think what happened was a sheriff is about I, don't know six months or more again, maybe now back in August hurt so somewhere around there Fedora basically bypassed the net socket layer entirely and just did all the parsing HTTP parsing in C++ land, and so that's why I didn't it didn't update that Valley after that.

G

We needed an option to drop that method like is that not something that people depend on or.

G

If nobody noticed so far, but well.

H

It was noticed because Sam Roberts Sam github currently was going to do some either either test tests or being able to do some metrics based on it and discover that it was always zero. So we have an open issue. You know, which is what raised it. You know brought it to our attention, so there are folks that we're trying to use it. Maybe.

J

H

But there are some, and it is something we do say that incoming message is a readable stream instance, and we say that readable stream has, despite read right, so we're kind of violating the contract there, based on what what we currently document.

G

And I guess from what you said the way it worked before right.

J

Yeah billig wings, me too, oh just look at this again, I mean how many people other than feud or how many people have actually looked at this and tried to find a solution to it.

H

I could probably find a solution, but it would take me a lot longer than fedora. Oh yeah well, I mean.

J

I'd even be willing to it kind of bypasses the GBS streams the c++ bypass, but I mean you can even do this in user.

C

J

Setting up a connection listener on the HTTP server and then setting up the data event on the on that connection and then just count the bytes going through. So it's possible right question is what will be the result of it and yeah. Until we a little bit more of an investigation, I say we just don't bother discussing whether we drop it or not, because it's a right.

G

It probably should be tagged. Us needs to be resolved before six goes out, though, because like if it's, if it's okay in previous release, probably want to make that decision before then right.

K

See I thought that the deal was you couldn't even get data events anymore, just because it was all being consumed at the C++ layer.

J

Know the we look up, you can't get readable events, you can still get data of them right. So if you do readable and then you do try and you doc data dot read, nothing shows up there. But if you listen to the rug like just on date, event it'll still pass it through.

K

B

Whoever asked a bit for six point, oh, that it looks like it's been happening since no just 4.1.

G

B

G

Just not it too.

B

Good, a while already yeah.

G

There won't be a surprise when people move to sex for that. Okay, I agree: it's not a walker on that front.

B

Do we have anything else to discuss a lot.

B

There's no I did actually look up. The MPN license thing so the license in v4 point ax is actually updated in mpm, but it's possible that it has not been updated. Our combined node license I. Think rod was talking about a lot.

C

Easy to do dis, ting of them I linked through the line stitch. Bothering me.

B

Where do you like to it, but I must.

C

Initial Angie in the church, I think I.

A

Can only say to them right.

B

um So that how it is, is that still is there I just.

C

C

But I just see it is in the link yeah to.

B

The church and I see that now so that is under there's a there's, a heading above that that says the following additional terms should apply the use of the the MPN cell for the NPM website and AM PM repository, but that doesn't actually, as far as I understand that doesn't actually apply to the code, because it's the license file no longer says it's um additional terms on the artistic license. So it's a pretty sure are our legal. Whatever.

B

Actually, okay, you'd say that it's.

A

C

To the FM software, the land turn your intent. 30 H says this: the phone additional dollars shall apply to you so then I'm software them and website, because Jordan vania, the services of Berkshire's obeying them and the lies that I have to say their traditional policy-related draw any distinctions on user from p.m. products and.

H

C

That I think includes the nfm software, our bond Olympian website. All such policies and restrictions as updated from time to time are hereby incorporated into this license agreement by using NPM. You can always your image to all such policies in the descriptions and NPM is just the Clint, the client.

C

If someone is using in them, it's not necessary that he or she is using then pimp depository.

B

Right, I think the terms only I'm like I we're not lawyers, so I'm not exactly sure how it works out, but I'm pretty sure that legally, it only works out that the dot actually applies to the end game as a company and npm like their services and the actual client is licensed under art. Just the artistic license. If there's further issues with that, we should I, don't know, make sure there's some issue for it and we should probably discuss it offline because we're not really more a very nice meeting. Well.

J

I mean: can we.

B

J

Discussing it, we have it, we have lawyers as part of the foundation when we just yeah.

B

J

B

J

C

Want to note that the license in the new branch wishes.

A

C

3 point something is fixed and it doesn't help that close anymore, and I just don't see why not too late the wisest dog branch.

B

Will have to take a look at it afterwards, um so moving on.

B

Stephen you had something about checking this small I, see you into the repo I. Think we've had this discussion before. What what are you proposing exactly yeah.

I

So the so this was given a month having this one check this you take this they're.

J

I

Five days ago, so I think I. As far as what 11 question is just so where we are at the latest I kind of gave an executive summary into.

G

I

Team here in the in the agenda, this includes a bump to ICU p57, there's platform issues, currently one of them frisbee previously just a Coast versus target tool set command line issue. This is not a new issue with this, but it's just exposed. So this is obviously been on the two issues about to have been longtime issues with these platforms, but we haven't seen them once I see you for some reason um right, but let's see here so so.

I

The chip issue, yohannes, I think, is- is fixing that or has a fix in progress basically and the raspberry pi cross build. I I need some help with how did how to actually do this cross. Build and I can try to make some progress here, but um otherwise this should be I mean this should be pronounced, ready to go now. I might be able to improve the build time, someone that could be a future change to this and.

C

I

I might be able to reduce the source files further and just as far as number of files checked in this is a question on what the with the right timing for lining this is.

B

Right so last time, I wasn't sure if we were actually going to go ahead with this in terms of crossville issues, you probably should talk to rod about Doc's I think he didn't most of that money has tied I. I'm stephen also millions here that this size would be about 20 megabytes in depth. / ICU dash, small, that's 892 files and the part of why I want to bring this up is because I think I saw some discussion of the James is if we should actually be building with I, see you by default.

B

H

um Yeah, there's a right right now in order to bility at the manually manually enable it, but our releases its on by default. There's also some that I actually need to confirm this. With the VA team and Ally.

H

There was a comment made last week at the BM summit, but there are the premise and discussions about whether or not to.

H

Rely more on icy, you NV, 8 and pull out or no longer supports some of the kind of the more homegrown unicode functions that are in VA. Can you confirm that I.

F

Ki yahan would be the right person to answer that one um I was just talking on the issue. Oh look at the time zone doesn't match for Germany what this discussion he.

H

Apparently, within v8 there are some unicode features that, if I see is not present, they provide an alternative implementation for, and the discussion last week was either that it was being looked at as either being considered or being planned to no longer support those additional pieces, so that, if I see is not present, dualities like spectacle see ya.

F

So the way it would work, for example, regular expressions rely on respective blank regular expressions acquire. I see you to be present. So if it's not present, your your javascript will behave differently. So so that's really was a concern like if node is building without I see you, you may get speck in compliant behavior, right, yeah,.

I

There are more and more features, especially with es6 that require ICU such as such as the regex, and also even string dot, normalize strained up. Look. How compare, I think is is in the form of stuff, that's four or two, but certainly string dot. Normalized is one that's my 262 compliance issues, so there are more and more things that are not just in the kind of Intel corner of things.

I

But there is, there is the there is. A desire to you know, reduce footprint on small, smaller platforms. You know footprint here is five megabytes of binary. At this point.

I

In English on the case.

B

Should is this discussion leading towards holding off on this and waiting for more discussion somewhere them all right.

I

I think the platform I think the platform issues are probably gating at this point and because otherwise, as things stand as I understand it and correct me, if I'm wrong the if the freebsd in our PI C is starts only over, that's that's a broken. That means were broken. So I right.

B

I would be yeah good how much how much compile time does ice you take is like. Another thing we go think about is like we already take like a really long time. If anyone wants to compile it on something like that,.

B

I

I yeah: that's why I put my note about in true bill time, because I realized I may be able to improve the building.

G

I

Further because the data I was just realizing this, as the cross-platform issue is going through since the data files already pre strips I, don't need to build as many of the tool tooling in this case, even compared to the choice you from download as we do today, because there's nowhere there's nowhere. You can download the exact data file node uses, but if we check in that data file- and we don't have to rebuild it so I can probably improve the build time.

I

A bit more I, don't have exact metrics on the difference, but I think I can get that for you.

B

Okay- and you said, there'd be updates to this probably four times here or something to ICU. No.

I

um I, probably I'd say max twice a year, but I mean no. No, it could pick it up once a year and probably be be fine. That's about what we were doing at this point and I don't have an idea of how many of these files would be would be changed. What the Delta would be. Definitely the what's it called the one megabyte binary file would be, would be a delta and some.

C

I

Other files would be.

B

Okay, I guess we'll take that back to the issue. Then.

B

James, you talk the next one, so this is dark and a TLS deprecated create secure pair.

B

H

um Yeah this I don't have a lot of context on. It was just a conversation that came up from coming tonight.

H

Yeah I, don't have the issue of in front of me, but I was proposing that this b.com deprecated and because of that it just needed to be reviewed, plus one that I believe we've talked about before.

H

It's really, you know. Are there any objections to dr. Karen deprecating? This is the reason to keep it around.

G

What's the rationale for deprecated, oh.

H

I, don't believe there was one given any oh I commit in the.

B

So the description is create secure pair uses, TLS underscore the legacy in the legacy connection, a connection class I think from node underscore crypto CC deprecated them in favor of using TLS, socket class.

H

Probably one that we're going to maybe half the door and yeah.

B

H

Anna, oh yeah, I.

G

Like the suggestion of an example saying here's how you use the new one too,.

B

It looks like chakras grep for its usage and mpm modules. Wasn't too large, so.

B

Then those libraries might be used to get fit, though I'm not sure. Ok, well, just move that too, when they're here.

B

The last issue was something I brought up so Brock some context on this back, like probably yeah. Back a year ago, we decided to deprecate the util. Dot is type matching functions because there were some issues necessarily maintaining those, and some people disagreed on certain things like is objects how that I'd actually function. So we went ahead into the docs deprecation and this was supposed to be a actual deprecation that was scheduled for six point out.

B

However, now we've actually changed so the parts that would be potentially broken by having object to string being modifiable, we've actually moved to C++, where they are virtually guaranteed to be correct. So we can def like we can guarantee that right. Exes are we're actually returning true for reg, X's and balls for anything else, because that Chuck has actually done C++ now and an issue is we don't really use these functions?

B

Necessarily we use some of them, but are they worth keeping around and exposing mostly because otherwise people would have to build a native add-on to necessarily check if you know something is for sure a symbol or whichever the ones used object to string like and eight does anyone have thoughts on that.

J

Okay, so if we're worried about allowing users to be able to use a function that allows them to find information out about something, and basically we want to say: okay, we don't want to you- want to allow a safe way in case anybody changes object to street. It kind of sounds like the argument. What would prevent anybody from overriding the utilities methods we'd have to make them all read-only.

J

Almost we're willing to make all these methods re only um a guy would be a at least a baseline of saying, k. Let's keep them.

B

um I, don't really have much to comment on.

J

The silence is sound, like nobody else knows either so seems. Nobody here at least has a strong opinion, whether we didn't refute them. I.

K

J

K

Would, rather than you know, keep I guess of the ones that are easy to do. We you know without having you know, to build an add-on, or you know whatever I think it's just it's much nicer for the end user, but I mean like the ones that are really simple to do in like JavaScript plan or whatever. Then you know I'm fine, with keeping those deprecated.

C

It looks like a nice.

G

Lady three chicks.

C

Function is peachy used in the 10 miles, a min that there looks like there are many models. Let's use it.

J

So I haven't looked at recently, but are we basically just taking all the v8 value? Is methods and exposing those to each other scripts and I'd basically be like a whole bunch of one-liners and C++.

B

There is a lot of one-liner things. A lot of a lot of it reduces down to type of checks. Like is string, is symbols. All those are is a type of check and I think they're only there for consistency or something we do also have again. The ones such as is date which are now done in C++, which are in C++ just a really easy type check and I'm not really sure why these checks don't exist just in JavaScript.

B

Maybe we should ask tc39 about that, or something there's also some combined checks that I think we originally used internally, like is null or undefined exists for some reason, and also we have in his primitive, which is sort of like, is no 9 defined mixed with some other check, or something like that.

J

Ok, so the dismal to find was at the time Furcal was me being kind of sarcastic to Isaac, because the reason these came about was that then put in macros to do quick checks and Isaac didn't like macros and code, and so he went ahead and swap them out with the is checks and I said. Well, let's go ahead and extend them with the full set. um Then the primitive simply it I put into actually a to implementation of primitive.

J

But the problem is that has changed with new spec and so honestly, the easiest way to do that at this point would be to just call in to C++ and Duty is primitive check. Oh.

B

So is that one we should change to c++ call, because it's not right now would that change.

J

B

J

Feeling a personality.

B

That's like the entire problem we had before, where is object, also returned the function. No.

B

Yeah also returns true for functions, and some people think that's incorrect. Okay,.

J

Well, I mean okay, so basically returns what the class inheritance model is, and so, if you're passing a strange day, is permanent sure yeah. It's going to return true if you pass in Rick's okay. So if you return like 50, but if you pass a new string to is object, then it should return through right.

J

If you just pass in a string literal, then it shouldn't return true, as far as as far as I'm concerned, whatever the C++ API returns is the correct implementation, and so if it changes functionality by switching to the C++ API, then ours was wrong.

B

So is the thing we're sort of settling on is to deprecate all the things, they're, just type out checks and only expose the C++ ones.

K

That sounds good to me. Yep.

G

What was the original reason suggested to deprecate them, because some of them were harder to implement than others and work, maybe keeping the harder ones and getting rid of the rest.

B

So I've already gone over a little bit. It was sort of a mix of these weren't necessarily used by us and they can be done easily in user land. Also, some of them, like is object, are a little bit controversial in how they necessarily work, because you could say, or a function as an object or a function should not show up as an object, because we're only testing object, literals and that's a little bit unclear and I'm, not sure there really is a way to clarify it by using whatever be AIT's. Api is yeah.

J

Like in the function case, you'd have to say: okay: well, are we going to follow the prototype chain in case yes, you're going to hit object, but as far as like a class inheritance model function does not inherit from object right, and so in that case it wouldn't be, but yeah it really depends on which avenue you say we should take in in type checking right.

B

The other there was one last reason, and that's previously for some things like errors and dates which don't have native type of checks in JavaScript, we used object to string to see what it, what they, basically what the constructor name was, but that is going to be modifiable in es6 now, so we can't necessarily guarantee that and that's why well that's why some of those checks have been moved to javascript now but sorry to C++ now, but originally that was a reason for doing this.

B

Okay, well, um unless anyone has any other comments or objections or anything I'm going to update this to deprecate everything that can be done easily in JavaScript and then only have unduplicated the C++ links.

E

B

Okay, I think that's all for today we're already past the hour, so I think we should probably just close out if.

J

If I can just very important, there's no discussion around it's just to mention, so it's underweight lines, we, the module EP, we need to settle on we've already settled that an extension was the best idea, but we haven't settled on what that extension should be personally I think we should just stick with the jsm from all the discussion it seems like that would be the easiest to use, but we will need a formal vote for the EP so that we can so defy that and land it x is 0 is.

B

There any reason not to use gsm / something else, because.

J

Other people want different extensions like where.

C

Folks, where did you use a suggestion for their extensions.

J

C

Firefox already uses the gsm extension for the morals or uses up my zero firefox and those files are also present in some and fam models. For some reason, oh.

J

Okay, well, what some I'll put on the agenda has more discussion on this next week. We have it. This is exactly one. Is we hash out of what the ctc is fine saying? Yes, we will we're okay using this extension go and then it uses completely for next week.

E

Thanks everybody bye thanks.

J

Thank you like that.

K