Node.js Core Technical Committee, 5 Apr 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: 2017-04-05 Node.js Foundation Core Technical Committee (CTC) Meeting

Description

https://github.com/nodejs/CTC/issues/93

A

Live now welcome to the nodejs foundations, core technical committee meeting for april fifth 2017 and just like every time we're going to start with a stand-up I, don't think honest he or she not seeing her. Let's uh Bradley, it's not hear Ben's, not here, so we go to Nikita.

B

Yes, nothing worth mentioning for me this week and next one is gone. Yeah.

C

Nothing worth mentioning just issues and for quest. That's the Devon.

D

Ya bana trying to get the cryptid random Phil and Rama Phil sink PR wrapped up it's about it. Next is franziska. Nothing.

E

Of note, Nexus Jeremiah.

F

If there was anything I, don't remember what it was, I don't know who's next.

A

Better is next, but I don't think his microphone is working.

C

Test test: oh there.

A

We go it's working, okay,.

C

I just opened fire folks to fix it anyway, so I have been working mostly on duck jeep, it's part of jeep dodge chest to duct tape, uh so it's probably quite interesting project or not yet, because probably let us build novel itself without fightin, at least in many systems and without notes pizer. So it's kind of very good for but bootstrapping note on platforms to doesn't have tied me. Both says it's pretty much it for me. Next is josh.

G

Yeah, I just did a couple PRS related to debugger, stuff and note inspect stuff and yeah next is Michael Dawson. So.

H

Just setting up our the node foundation, no GS foundation, user for NPM management resolved a few buildings relais de IX, a couple of any P I Pierre's to fix Coverity issues reported at PR, landed working with the rest of the API team on documentation and starting to look at setting up some see I testing for the supporting modules so that we can test that as well and then just for miscellaneous, you know reviewing and landing things. Next is Brian. Wait.

E

And not too much for me other than the usual commenting on issues and reviewing PRS next is miles so.

I

I put together a meta issue for the remove of legacy debugger protocol, which we'll talk about a little bit, did some research into turbo fan and ignition released. These six point 10.2 and be four point. Eight point: two: there was a leak we talked about it last week, but I can give people more information, but if they want but also released, a new version was carrying the gold mine. Next up this alley, yeah.

J

I did some work on the debug / inspector. Be ours, that's about it. Next would be rich.

A

Hi I'm scheduling some onboarding for some collaborator, nominate people who num they'd be collaborators preparing for code and learn which will be in oakland on april twenty seconds. So if you are a collaborator or or even just a contributor to node core- and you would be interested in that- let me know our comment on the issue in the code on the repo and other than that. Let's get started, I don't think we missed anybody.

A

If I missed you in the stand up speak up a last meeting, we talked about buffer a lot and we had a series of votes on the in the github issue. Tracker.

A

The end result is that for version 800 of node, we intend to have an opt-in flag for runtime deprecation of the buffer constructor and we intend to 0 fill and we do not intend to backport that at this time at least I believe that covers all the all the bases on that, if I'm forgetting something, maybe someone will you might be okay, cool um and then so this week we have one item on the agenda and it's all about removing the legacy debugger inversion.

A

8020 of node and miles this is your issue, so do you want to take it away.

I

Sure so, at a very high level, what we've got going on here is right now run 5.7 on master and in eight point. Oh five point. Eight is actually going to be promoted to stable right around the same time that we're planning to cut the branch and in the past we have actually run. You know beta versions of v8 in our tree before a release. As long as it's stable before you know, we cut our release. So the challenge here is that a v8 5.8 there is no longer going to be.

I

You know any of the code or protocol related to the legacy debugger. So if we want to have a path to 5.8, either during the beta and RC process or in the life cycle of eight, we will have to remove all of the legacy debugger code. So this has been a bit of a meta issue which is collecting all of the things that will need to land.

I

Josh can come in and give or Michael a little bit of a heads up as to what was discussed in in the Diagnostics working group. They've already discussed this and voted on it, so they can talk to about it. As far as I know, my impression was that they were all positive and for the debugger being removed the you know what this would do. The things that we're talking about doing is aliasing debug to inspect.

I

We would have to move the v8 inspector to stable, because, right now it's not stable we'd have to remove hyphen hyphen G bug. We want to update the hyphen hyphen inspect hint text. There's one more issue here for under get under a score debug process to activate inspector sig. Us are I'm, not super aware of what this one is doing. That I believe that Allie can't ride in about about that. So I'm going to hand it off to Josh right now to just give an update from the Diagnostics working group.

I

If you're up for that Josh.

G

Sure, an ally you can chime in here essentially I decided and we should have put in an issue ourselves, I guess on the main thing that even if we go with 5.7, there's really no reason to keep the dash dash, debug, the old debugger interface or API even accessible, which would address what miles brought up that that way.

G

We even if we start with 5.7, we can still go to 5.8 um without making without being some for major, and that was that was that was the conclusion in our last meeting and allie has updated the the way we would hide away. The interface of 57 is what the alias that Ali is working on. I think so that's about the summary just.

I

A question for you, because I believe that something is true, but I want to make sure that it's that is true. Have we begun shipping a note inspect as a replacement, CLI debugger inside of our tree yeah.

G

In the latest version since about 76 or 77, it's been that moon so.

I

I just wanted to make that clear, because I know that after the Diagnostics working group, there was some misinformation on Twitter people thought that we were talking about removing this without any solution or any options as far as the CLI debugger is concerned, so for anyone who's listening and just for the record, you know if we move forward with this, we do have another CLI debugger in there, and the main thing here is: you know: removing a lot of deprecated code and making sure that we're not shipping stuff that blocks us from upgrading other things.

I

Richard's just asked in the comments really quickly is the usage of the bundled note inspect document today. Is anyone able to speak to that I.

G

Can check and remember.

I

Okay, I will add this to the list of things that we need to need to do. I think.

J

One of the PRS that get it gets rid of dash. Debug, I think, does make the updates to the docks so so.

H

Just in terms of thumb, like information on usage of the old one, do we have any any data on that.

I

I've got some fantastic anecdote data if you like that, it's not scientific, but I do know that for a non-trivial period of time from one promises landed on core to about a year to a year and a half after promises landed on core, if you called inspect- or you till, I think it was like 11 of a handful of different utility debugger would just explode and it took almost a year before anyone actually sent us any issue, issues about that and when we actually found it out, it was like another issue that led us to realizing that there was that weird edge case.

I

So the fact that the current that the legacy debugger is not very well tested to test themselves, don't really work, at least on four and six and the fact that it had been in production and broken for quite some time without any reports leads me to believe that the usage is fairly low.

I

Similarly, on an anecdotal level, I have talked quite a few people who say that they do rely on it once in a while, but I. You know a lot I've seen tons of stuff online and tons of you know, documentation around using the new inspect I API, doing introspection and debugging using Chrome tools or Firefox or Chrome I believe that Allie can speak more to this. But a number of the external vendors that were relying on the old debug API have all updated, so they're all using the inspect API.

I

Now my my gut is telling me that the number of people that this will directly affect will be fairly low, so that, uh yes.

J

So there's two things right, so there's the CLI, debugger and I think we have heard any ghosts. Very few people used it and ever it was not tested and in it yeah there were lots of issues and we were supporting on the contest record, pads and other thing. It means we have a better debugger now so that so any users who are using a CLI debugger will ever see a debugger.

J

It's going to be better well as for the dash dash debug protocol, so the end-users weren't, really users of this across the CLI, debugger and and and debugger vendors, such as web store or a node inspector. So so all of the vendors have pretty much migrated to the inspector protocol. Now I think node inspector is no longer be maintained, I, don't think it as users, anymore. People have moved over to chrome, dev tools or big estudio code or or web from another debuggers, which are all now using the new sort of power.

J

So it's wrongly, if there's any real users left for the old JSON based developer forum,.

G

Yeah, we certainly haven't heard anyone making a big deal about it. Nobody said it needs to stay or give us another version. Yeah.

A

I'm of the opinion that the sooner we pull off the band-aid, the better.

H

A

K

A

H

K

H

Probably the most important thing say.

I

So, michael is a question to you. So note inspect was a strongly freako. The last time I checked is there? Do you know if there's any plans internally at IBM, to continue maintaining that yeah.

H

I think I think I saw an internal discussion that says hey, you know when, when the new inspect, when the new protocol comes to be that's, you know it's it's going to be obsolete and we're going to let it be that way at least for now so I just noticed. It says that on there yeah it was.

C

I sated to say.

H

It's not maintained I believe so.

H

I thought, if all the other vendors have changed it, that one I, don't think, is a concern.

I

So with that, with that being considered, we right now have 4 9 15 members of the CTC on this fall or 14 members and an observer.

I

Is there anyone who would vote no on this or, alternatively abstain? So if we have neither of those that I believe that we have forum on moving forward and coming into the book, yeah.

A

I mean just to get consensus. Are there any objections to moving forward with us.

A

Sounds like you can proceed excellent.

I

Thank you very much. Everyone I really appreciate the time retro hand it back to you.

A

Prepare for that, but that's okay, I'm going to get prepared really quickly! I mean I! Guess it's QA times so I'm going to slowly read off the upcoming meetings. While we try to gather questions from the usual channels on IRC and Twitter and where YouTube than wherever else here.

K

Real quick here so for one more thing: what.

A

K

See a ghost here, wanna do.

K

But I'll me out so many more reviews for that tool and I would I. Don't doubt it's going to get in 80, but I think it could land in eight dot one. There are really two aspects to it that need to be reviewed. One is to see if it pour as it currently stands. So there are no regressions and then the second aspect is for add.

K

Operating they need to be.

I

C

Up quite a bit.

K

I

Miss that last sentence, but.

A

K

A

I try to summarize real quick.

K

It needs to be reviewed first land I'd like it to land in 81. It needs to be reviewed in in terms of not affecting Kourt, so no regressions and also to make sure that the API as documented does work correctly.

K

So anybody who feels like reviewing something feel free, I, don't even need for reviews. It's a big PR, don't feel like you need to go over all of it.

A

Awesome so your main your main obstacle right now is that not enough people are reviewing that PR. Is that fair to say, yeah.

K

A

So let's get some reviewers on that cool anything else about a sink hooks from Trevor or anybody else.

F

Awesome I would like to make a comment about the debug API, considering we are removing it without having and LTS in between I. Don't exactly approve of this, but I'm not going to object to it either. So.

I

Jeremiah I.

F

Don't feel like perfectly comfortable about it, let's just say.

I

Jeremiah I, don't know if you recall, but we did backport the deprecation warning to set it for this exact reason.

F

I

F

There I mean there's people that are using seven, but it's not like most of our users.

I

um So we could I mean we could theoretically do a minor of six but I guess we don't have enough to be time to do a minor six. That would include a deprecation warning. I guess.

C

That deprecation warning in six.

I

Would find at the assembler major? So maybe it's a it's a non-starter there. We.

G

Did post a blog, post and stuff about this I? Don't know if that helps. We did trying our best to get the word out, and so anybody has any problems they can. Let us know so, for nobody said anything yeah.

A

I mean the alternative, is we don't I mean the only alternative? Is don't upgrade v8 in 800 and I? Don't think any of us want that.

F

Like I, don't really really have another like a great other option, I, just not the biggest fan of doing that removing things but I LTS is in between, since that's kind of what we've landed on usually doing so just.

I

F

That was under.

I

The impression that Lee I was under the impression that it was more just that we had a major with a warning before we did it in the next major is it is there other examples of where we did this with an LTS in between or like stalled due to an OTS in between looking.

H

Normally you'd have a deprecation cycle, wouldn't you like you deprecated? It there'd be at least a release. In this case, this the six release, the LTS release, at least you know, wasn't deprecated there. It's just going to disappear in the next one. I think.

J

That applies to no DP is so. If there's a v8 dependency or a live Yuri dependency that we depend upon and then we can't have, we don't offer the same deprecation cycle guarantees for those api.

H

Yeah thought: that's not a good thing, necessarily it's just the facts of life right. It doesn't mean that, like in this case, we actually have a choice. Often we don't, in which case we don't have a choice. So you know we can't do it, but.

I

But we also have the choice to like remove code that we know is buggy and then we know is unsupported rather than supporting it for three years. So to me, it seems like a less risky option to actually remove it than it would be to keep.

H

If the key thing is whether we know I think the earlier discussion was, is we don't believe anybody's going to be using it right, mm-hmm, that's what it would come down to the you know if nobody's using it, then getting rid of the code is a win if people are using a thing yeah, it's fair to say that you know it's a little bit of an uncomfortable situation to say we're getting rid of it without the deprecation cycle.

H

If 57, for example, didn't have it in it, we probably just I, guess, say: well, you know it's out of our control. We removed it.

I

Well, I mean we could go forward with also, like you know, removing this stuff doing calls to the community, and if we have any responses from individuals that are uncomfortable with what's happened, you know a reversion to put it back in can always be on the table before the RC.

A

I mean there's precedent for reverting breaking changes. Even after the release, I mean we did that with the runtime deprecation for buffer constructor without new it.

H

Ought to be here when he's sober minor, to revert it too right, yeah.

A

Yeah the issue that the potential should we run into is: if we upgrade v8 to where you know, we can't support the debugger anymore, but I mean I again: I, don't I I, just don't think. Really we have any choice here. I think we have to get rid of it and I. Think yeah I mean like it's not the greatest. You know it's not the greatest circumstances.

A

It would be great to be doing this, nay, you know more controlled fashion, which will sound a lot more control, but in the fashion that you know, took more releases and gave people a lot more warning, blah blah blah blah blah, but we've you know, we've we've done everything we can, and you know the you know, I think I think this is the way we have to go. I. Don't really think unreal. I think there's another viable option. I think.

H

We just need to keep our ears open, for maybe what it was suggested is, you know, keep may be publicized as best. We can and then keep our ears open for if it is causing grief and hopefully the answer is it won't and no they're. Just there. Okay.

A

All right, unless someone wants to add something more, it's time for upcoming meetings and I know if somebody else wants to read these items, mess them up, but we have a collaborator summit for node.js collaborators in Berlin on May. Fourth and fifth, there is a tc39 meeting which I I'm not sure miles like this. This kid who's who's invited to be to this is this for ctc only or will who's this. This may twenty second thing for this.

I

Is for Casey, 39, CT, CTS, see and board members and members of the foundation I've gotten to extend it to full who are involved. You know with node and you know, would benefit from meeting and greeting with the TC 30. Ok,.

A

I

A

I

A

God, please go for it. So if you're, one of the few we like we have like you, know hundreds of individual members of the foundation so.

I

There are this isn't for individual members is for CT CT CTS, see when I say members of the Foundation's I mean staff of the foundation. Gotcha, ok,.

A

Ok, so this this is this is this: is for few dozen people, um ok cool, so that I'm just going to move on because CTC meeting next one is a week from today, but much earlier at oh-five-hundred UTC cutting on what time zone you're in that will be that you have that wrong, not.

G

Sure, if it's really 500 UT is it yes.

A

It is I, think I think I put in there. I could therefore it can't possibly be wrong. Nope just kidding. um Well, let's check. I think it is 500 UTC, but we might as well take the 10 seconds. It will take to confirm.

A

Yes, five five o'clock UT at five a.m. UTC; yes, that's correct and then the tsc meeting, the next one is tomorrow. This says at the same time, this meeting started 20 hundred UTC 8 p.m. UTC um and that's what we got do we have any questions lined up.

F

Yes, someone asks flow there. How can I contribute to notice source code I'm a front-end developer and I concern myself, a good jazz developer and note any tips around? Is it possible? Oh? Can.

A

I take this was possible, but the seminoles can elaborate more. My recommendation would be to go to node to do n 0 de Tod 0, 0 RG, and there are instructions there for getting help and getting started.

A

Additionally, the foundation periodically holds events called code and learn which are specifically about getting people involved in contributing to their first contributions into nodejs core, so be on the lookout for those events, there's one in Oakland California later this month and then we're looking to have later this year. If all goes according to plan one in China, one in Ireland and one in Vancouver, Canada.

A

A

Any other questions.

A

A

Someone else want to say, but oh god I just.

I

Gonna say I don't think we have any.

A

All right cool, ah so oh wait! Glad you're mine is.

F

The one in vancouver at the same time as know Jason tracker, yes,.

A

And the one in Ireland will will hopefully be at the same time as node compy UI, and the one in China will hopefully be at the same time as jess comp CN, um but the only one that's definite right now is the note interactive one in Vancouver and.

F

Just to clarify at the same time means I think either the day before or the day after or something right.

A

And probably a very good chance, it won't be during the event. It might be the next day or the day before, but it will be around that time.

H

The gist come for Berlin. Is there one happening around that time or not? No.

A

There is not okay.

H

Good just want to be sure.

F

F

A

All right: well, thanks for listening everybody we're going to move on to a little private business. So that's it thanks and next week.

D