diff --git a/meetings/2016-11-16.md b/meetings/2016-11-16.md new file mode 100644 index 0000000..107bc2b --- /dev/null +++ b/meetings/2016-11-16.md @@ -0,0 +1,242 @@ +# Node Foundation CTC Meeting 2016-11-16 + +## Links + +* **Audio Recording**: +* **GitHub Issue**: [CTC#33](https://github.com/nodejs/CTC/issues/33) +* **Minutes Google Doc**: + +* Previous Minutes Google Doc: + + +## Present + +* Anna Henningsen @addaleax (CTC) +* Ben Noordhuis @bnoordhuis (CTC) +* Сковорода Никита Андреевич @ChALkeR (CTC) +* Colin Ihrig @cjihrig (CTC) +* Evan Lucas @evanlucas (CTC) +* Jeremiah Senkpiel @Fishrock123 (CTC) +* James Snell @jasnell (CTC) +* Josh Gavant @joshgav (observer/Microsoft) +* Michael Dawson @mhdawson (CTC) +* Brian White @mscdex (CTC) +* Ali Ijaz Sheikh @ofrobots (CTC) +* Jenn Turner @renrutnnej (observer/Node.js Foundation) +* Michaël Zasso @targos (observer) +* Myles Borins @TheAlphaNerd (CTC) +* Trevor Norris @trevnorris (CTC) +* Rich Trott @Trott (CTC) + + +## Standup + +* Anna Henningsen @addaleax (CTC) + * Not much, catching up on things +* Ben Noordhuis @bnoordhuis (CTC) + * Same old; code review, fixing bugs, first-line support for my sins. +* Сковорода Никита Андреевич @ChALkeR (CTC) + * Some security investigation stuff in the ecosystem, nothing else to report +* Colin Ihrig @cjihrig (CTC) + * Reviewing issues and PRs. libuv 1.10.1 release. +* Evan Lucas @evanlucas (CTC) + * small PR, some comments on PRs/issues, not much else +* Jeremiah Senkpiel @Fishrock123 (CTC) + * nextTick & process warnings code coverage (coverage.nodejs.org) + * misc prs & issues + * bot script to attempt pr backports now live, automatically adds some labels + (https://github.com/nodejs/github-bot/pull/90) +* James Snell @jasnell (CTC) + * nodefest.jp + * HTTP/2 +* Josh Gavant @joshgav (observer/Microsoft) + * reviewing tools and libraries for V8 inspector + * version management group (https://github.com/nodejs/version-management) +* Michael Dawson @mhdawson (CTC) + * Aix resizing + * validation of updates to coverage job + * keeping up with issues and commenting + * tested out config so we can get emails from jenkins + on job failure + * scheduling next benchmarking meeting, issue to discuss pruning + Membership + * Ongoing work on ABI stable PoC + * LTS/Diagnostics meeting + * Misc reviews/lands +* Brian White @mscdex (CTC) + * Reviewed PRs, replied to issues +* Ali Ijaz Sheikh @ofrobots (CTC) + * Nothing major to report +* Jenn Turner @renrutnnej (observer/Node.js Foundation) + * No update, just observing +* Michaël Zasso @targos (observer) + * Working on ABI compatibility patch for V8 5.5. + * Reviewed some PRs. +* Myles Borins @TheAlphaNerd (CTC) + * Backporting to v4 and v6 + * Mentoring new potential backporter @gibfahn + * Mentoring new citgm collaborator @GeorgeAdams95 + * General review of issues + PR’s + * Participated in initial version management meeting +* Trevor Norris @trevnorris (CTC) + * Fix perf regression after rebasing onto V8 v5.4 +* Rich Trott @Trott (CTC) + * CTC scheduling options, my head hurts + * The usual test fixes, Jenkins observing, PR and issue commenting + +--- + +## Agenda + +### nodejs/CTC + +* A new New CTC Meeting Schedule Proposal + [#29](https://github.com/nodejs/CTC/issues/29) +* Documentation WG Status + [#32](https://github.com/nodejs/CTC/issues/32) + +### nodejs/node + +* Revert "buffer: runtime deprecation of calling Buffer without new" + [#9529](https://github.com/nodejs/node/pull/9529) +* tls: fix leak of WriteWrap+TLSWrap + combination[#9626](https://github.com/nodejs/node/pull/9626) +* src: don't call into VM from AsyncWrap + destructor[#9467](https://github.com/nodejs/node/pull/9467) + +--- + +## Previous Meeting Review + +* Roadmap WG status + [#16](https://github.com/nodejs/CTC/issues/16) +* buffer: hard-deprecate calling Buffer without new + [#8169](https://github.com/nodejs/node/pull/8169) + +--- + +## Minutes + +### A new New CTC Meeting Schedule Proposal [#29](https://github.com/nodejs/CTC/issues/29) + +@trott: Propose continuing with current 2 times in the meantime, discuss with +Ben and others how to improve in the future. + +CTC repo (https://github.com/nodejs/ctc) has a list of upcoming meetings through +end of December: https://github.com/nodejs/ctc#meeting-schedule + +@joshgav: William Kapke maintains a Google calendar: + + +--- + +### Documentation WG Status [#32](https://github.com/nodejs/CTC/issues/32) + +@trott: William Kapke asked that people chime in on this topic. + +--- + +### Revert "buffer: runtime deprecation of calling Buffer without new" [#9529](https://github.com/nodejs/node/pull/9529) + +@trott: To be moved to private discussion. + +--- + +### tls: fix leak of WriteWrap+TLSWrap combination [#9626](https://github.com/nodejs/node/pull/9626) + +@bnoordhuis: I believe Fedor wants to expedite backporting this into v6.x and +v4.x. + +@thealphanerd: I suggested getting it into v7.x first, then prepare for LTS. +Fedor and others approved on IRC, so no changes to release plans. + +I staged that change in v7.x-staging and will coordinate with whoever does the +next v7.x release. + +I’d prefer if we can get a v7.x release by next Tuesday (11/22) so that there +are two weeks for it to bake before it lands in LTS releases. + +@Fishrock123: I’ll get a v7.x release out early next week. + +Next steps: + +* Fix in next v7.x release than backport. + +--- + +### src: don't call into VM from AsyncWrap destructor [#9467](https://github.com/nodejs/node/pull/9467) + +@bnoordhuis: Due to way V8 GC works code cannot call into VM during a GC. +async_hooks includes a `destroy` hook which runs when these objects are +garbage-collected. That’s never been safe, and is now causing crashes. + +Stop-gap measure is to remove the `destroy` hook for now. If someone wants to +work on an alternative that would be welcome. + +@mhdawson: Will removing it break stuff? Should we just have it do nothing (make +it a no-op)? + +@trevnorris: Previously we had a lot of weak objects so that callers wouldn’t +crash if the object was GC’ed (?). That’s no longer necessary cause we have more +guards. + +Several options to solve this issue: + +Remove all weak handles completely. +Use second-pass GC callback. +Put `destroy` callbacks into a uv_idle_t and let callbacks be called in next +loop with `setImmediate`. + +@Fishrock123: async_hooks is an “experimental” API, so technically it could be +removed or changed at any time. So we could remove that to make it not crash. Is +that a good idea? I’d prefer to make them no-ops instead. + +@trevnorris: That (no-ops) would be more dangerous cause users might be +expecting it to run and rely on it to clean up stuff. + +@Fishrock123: Well the alternative is that they’ll crash anyway. + +@trevnorris: Many apps don’t crash and they’d be fine. + +@bnoordhuis: Removing all weak handles would be a good option. + +@trevnorris: Will work on that. + +@trott: Do we have agreement? + +@bnoordhuis: Question is time. If the better fix won’t be ready in 1-2 weeks I’d +still push for removing `destroy` hook for now. + +@trevnorris: Will start on removing weak handles. If that’s not feasible to get +done before next release, will use the uv_idle_t approach and get done before +next release. + +@Fishrock123: Does it only affect those using async_wrap? + +@bnoordhuis: Yes, but some use async_wrap through second-level dependencies. + +@trevnorris: Also affects users of `node-weak` module for the same reasons. + +From YouTube: Should experimental features be kept out of releases? + +@thealphanerd: This is how we test new features. + +Next steps: + +* @trevnorris implement steps above. + +--- + +## Q/A on public channels + +None. + +--- + +## Upcoming Meetings + +* CTC: 2016-Nov-23, 1600 UTC +* TSC: 2016-Nov-17, 12pm Pacific +* Diagnostics: December +* Benchmarking: 2016-Nov-18, 4PM Eastern Standard Time +