8.8 KiB

Raw Blame History

Node Foundation CTC Meeting 2016-02-10

Present

James Snell (CTC)
Trevor Norris (CTC)
Colin Ihrig (CTC)
Brian White (CTC)
Alexis Campailla (CTC)
Bert Belder (CTC)
Chris Dickinson (CTC)
Shigeki Ohtsu (CTC)
Steven Loomis (observer)
Mikeal Rogers (observer)
Fedor Indutny (CTC)
Jeremiah Senkpiel (CTC)
Rod Vagg (CTC)
Ben Noordhuis (CTC)
Nikita Skovoroda (observer)
Ali Sheikh (observer)
Evan Lucas (observer)
Rich Trott (observer)
Michael Dawson (observer)

Agenda

Extracted from ctc-agenda labelled issues and pull requests from the nodejs org prior to the meeting.

Revert "fs: deprecate fs.read's string interface" #5163 & fs: add a temporary fix for re-evaluation support #5102
buffer: add Buffer.from(), Buffer.alloc() and Buffer.allocUnsafe(), soft-deprecate Buffer(num) #4682 & Buffer(number) is unsafe #4660
CTC Membership Nominations #4750
Seek legal advice on LICENSE and copyright blocks in code #3979

Standup

Brian White: Working more on improving performance in various areas of core, reviewing PRs and issues.
Chris Dickinson: Promises PR + discussions with the error symposium group.
Rich Trott: addressing flaky tests, enhancing linting, working to find a path to more automation in landing of PRs (although others, including Alexis, seem to have picked that ball up, thankfully)
James Snell: Working on express stuff in, security stuff out, buffer API finished, moving forward on string externalization.
Jeremiah Senkpiel: Hook for unhandled rejections that detects when the GC fires on a promise. That is currently blocked on V8 bug with weak callbacks + promises.
Chris Dickinson: Promises PR + discussions with the error symposium group.
Trevor Norris — MakeCallback reentrant fix for HTTP parser / AsyncWrap interaction.
Michael Dawson - Working on getting AIX up in the CI. Also on running v8 tests in the Node tree. Added AIX to libuv tests.
Steven R. Loomis - not much to add
Nikita Skovoroda — like ususal, mostly comments and the initial version of the codesearch API on a VPS. Nothing major.
Rod Vagg - Security stuff, catching up on issues and discussions. Legal committee meeting yesterday.

Review of last meeting

Enable Node.js to run with Microsoft's ChakraCore engine #4765
CTC Membership Nominations #4750
buffer: add Buffer.from(), Buffer.alloc() and Buffer.allocUnsafe(), soft-deprecate Buffer(num) #4682 & Buffer(number) is unsafe #4660
Seek legal advice on LICENSE and copyright blocks in code #3979
@srl295: path for full-icu data? - ./node_modules vs ./node/ ? (npm linkage) - todo, update ticket with meeting resolution

Minutes

Revert "fs: deprecate fs.read's string interface" #5163 & fs: add a temporary fix for re-evaluation support #5102

Rod: Simple deprecation warning for the particular argument usage. Pulling in the internal util module for deprecation messes up older versions of graceful-fs (and thus older versions of npm.) The question is whether to revert the PR, or to add a helper (that Nikita proposed.) I would suggest that this is urgent enough that we need to do something about it.

Jeremiah: This is a poor reason to revert it. James: From everything that I’ve seen, adding the warning is the least bad option. Folks seems to be against reverting. Maybe making the warning less specific to graceful-fs.

Ben: I don’t agree. I think it’s right to point them to the newer version.

Jeremiah: I agree (with Ben).

Rod: Anytime you break the build, it’s an immediate candidate for reversion. It’s not permanently reverted, but it’s “let’s fix the build.”

Jeremiah: This doesn’t actually break CI though. It breaks newer versions of npm on master. .. It might be npm in general just on the master branch.

James: Forrest noted that pretty much any change other than reverting will have a significant effect on the npm user base.

Jeremiah: Yeah but we’re not shipping this in a stable.

Michael: If it breaks in master, doesn’t that prevent folks from doing other tests?

Ben: I think the problem lies with npm here.

Mikeal: You make that sound so easy.

Ben: The version of npm we’re talking about is still in a PR, yes?

Jeremiah: We don’t regularly run tests on master, so it might already be broken. I can test it right now..

Rod: I think Myles has been trying to run the npm tests more regularly, as part of smoke testing — and it’s broken for the smoke testing now.

Michael: It seems like we’d want to keep things as green as possible. It seems like we’d want npm to do something.

Rod: It’s a fact of life that we exist in an ecosystem that has these kinds of issues. My vote is to revert, and then revisit after we’ve gotten the build fixed.

Ben: Well, yeah, I do. I think Nikita’s PR is an acceptable intermediate solution.

Jeremiah: Likewise. The resolution is that we try to require it, if that fails, we use a deprecation helper we build in that emits a warning that something is breaking an internal module — that would fix it and seems quite reasonable. It will also help people note where stuff is going to break.

Rod: Does someone want to help Nikita get this in?

Ben: Didn’t you already get a couple of LGTMs?

James: I think there’s one -1 and three LGTMs.

Rod: We can force it if we have to.

James: I propose: if we accept this now, let’s try to get it resolved better before v6 goes out — which gives us a deadline so it’s not pushed off indefinitely. npm gets a fix, and that’s where it happens.

Ben: I think it’s a good idea to push npm to update their dependencies. Whether or not npm updates, I think it’s a good idea to include Nikita’s PR.

Jeremiah: I think so too.

Rod: Let’s take it back to GitHub.

buffer: add Buffer.from(), Buffer.alloc() and Buffer.allocUnsafe(), soft-deprecate Buffer(num) #4682 & Buffer(number) is unsafe #4660

James: There were some changes I was waiting for from Trevor w/r/t fill encoding. There was a desire to bikeshed a bit more on the name, not sure if you want to do that here. I think absent that the PR is probably in a place where we can land it.

Rod: Sounds like we need a last call for objections, otherwise it’s going in.

James: Yep.

Rod: How about you do that on GitHub?

James: OK.

Jeremiah: Are you adding a zero-fill flag?

James: Yes — it will switch all allocations to use calloc under the hood. The one thing this does not do is change the default behavior of new Buffer(size). If we did that, it would be a breaking change going back to 0.10.

Bert: Is the plan to deprecate it off the table?

James: It’s a soft deprecate — docs only.

Jeremiah: We should try to just deprecate it. The problem is that folks are running different versions against modules, and it would be bad to rely on zero-fill behavior where it doesn’t exist [CD — didn’t capture this entirely]

James: We need to refine our deprecation strategy — and define whether something is deprecated vs. end-of-life. Will probably be returning to that next week or the week after.

Rod: It sounds like we’ve got a way forward, here.

Bert: It’s ugly but I won’t object.

Rod: Noted.

CTC Membership Nominations #4750

Rod: We should start to line up a vote.

Seek legal advice on LICENSE and copyright blocks in code #3979

Rod: There was a PR that Mikeal put in that updates the DCO from v1.0 to v1.1. If you’d like to review that, please comment on that.

Other business

James noted that the express application landed today. I will be working to move that into the new organization. More of an update on tomorrow’s TSC call.

Jeremiah notes that he will be a bit preoccupied by that for a while.

Next Meeting

2016-02-17

8.8 KiB Raw Blame History