You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 

8.1 KiB

Node Foundation TSC Meeting 2015-07-08

Agenda

Extracted from tsc-agenda labelled issues and pull requests prior to meeting.

nodejs/io.js

  • Default Unhandled Rejection Detection Behavior #830

joyent/node

  • Adding a "mentor-available" label #25618

Minutes

Present

  • Mikeal Rogers
  • Colin Ihrig (TSC)
  • Ben Noordhuis (TSC)
  • James Snell (TSC)
  • Fedor Indutny (TSC)
  • Bert Belder (TSC)
  • Michael Dawson (TSC)
  • Steven R Loomis (TSC)
  • Alexis Campailla (TSC)
  • Jeremiah Senkpiel (TSC)
  • Julien Gilli (TSC)
  • Chris Dickinson (TSC)
  • Shigeki Ohtsu (TSC)
  • Trevor Norris (TSC)
  • Domenic Denicola
  • Brian White (TSC)
  • Rod Vagg (TSC)

Review of the previous meeting

  • Policy for PR blocking? #2078
  • Resolution was to deal with it on a case-by-case basis for now.
  • Internationalization WG (Steven)
  • Steven Loomis is going to kick off the working group.
  • Steven: no further responses on the github issue.
  • James: just need to get started
  • lts: strawman LTS cycle lts#13 / Proposal: Release Process #1997

Standup:

  • Mikeal Rogers: wrote a new confrence call tool for us that uses Twillio
  • Colin Ihrig: Not much, reviewing PRs, triaging issues.
  • Ben Noordhuis: reviewed a lot of PRs, upgraded v8 in next and next+1.
  • James Snell: Working on the LTS Proposal, triaging issues in joyent/node, investigating stuff for the upcoming openssl fix.
  • Fedor Indutny: fixed node after v8 upgrade. Exposed critical issues.
  • Bert Belder: Not much code, had conversations with Mike Dolan and James Snell about the foundation and organizational issues. Working through a laundry list of libuv PRs blocking the next release.
  • Michael Dawson: Working on getting PowerPC to build on io.js, tested the security fix from last week, joyent/node triage.
  • Steven R Loomis: Worked a bit on the Intl WG, not much else.
  • Alexis Campailla: converged CI, almost done. Dealing with windows installer issues. Expect converged CI to work in a week.
  • Jeremiah Senkpiel: General triaging and reviewing, helped do the release last friday. _unrefActive with optimizations with heap timers. At CascadiaJS the next of the week to get people’s feedback.
  • Julien Gilli: Released 0.12.6 last week, working on setting up other people to do joyent/node releases, joyent/node issue triage
  • Chris Dickinson: Working on docs more, have a new tool for docs to make sure the links are correct in a tree of docs, started a collaborator check-in on the io.js issue tracker, hopefully will be weekly. Jeremiah: what is that doctool? Chris: “count-docula”, a MDAST-based tool to verify correctness of the docs.
  • Shigeki Ohtsu: Not much on io.js, preparing to update OpenSSL tonight to get the OpenSSL security fix out.
  • Trevor Norris: Investigating the UTF8 decoder security issue and working on the fix. Reviewing PRs and being involved in the W3C Web Assembly working group.
  • Domenic Denicola: Not much on io.js, travelling, stress testing the vm module.
  • Brian White: Triaging issues, working on the javascript http parser more & benchmarking it.
  • Rod Vagg: We should discuss the LTS proposal again since there was lots of work done on that. Working on lots, including the security fix from last friday (writing up a post-mortem for it), getting external people involved to review our security processes.

Default Unhandled Rejection Detection Behavior #830

  • Domenic: let’s say there was a magic way to detect when an error in an err-back style callback was not handled, what would we do? Print to stderr?
  • Bert: We do have a history of printing things to stderr. We should follow browser semantics if we can, in favor of primnting a warning but nothing else.
  • Discussion about the technicalities of handling unhandledRejections
  • Rod: not sure we should do anything since detecting this is somewhat arbitrary.
  • Domenic: there is a proposal for this that chrome implements behind a flag that comes close to how the unhandledRejection hook in node works
  • Discussion about the technicalities of having a better hook for printing a warning after garbage collection of an unhandled rejection.
  • See this thread for background detail of options in v8: https://code.google.com/p/v8/issues/detail?id=3093#c1
  • Action: nothing now, maybe if v8 adds a hook for when rejections get garbage collected.
  • Domenic: looking at v8, it seems to have most of the hooks, so this may be possible soon.

Adding a "mentor-available" label #25618

  • Folks are interested in contributing to larger tasks, need mentors to help them understand the process. Should we add a label?
  • Julien: Many people are interested in making “deeper” contributions, but they need a mentor. Let people add a mentor-available tag so they can locate these.
  • … part of the discussion missing here ...
  • Resolution: let’s try it, one such label has already been added.

Having more people managing releases for Node.js v0.10.x and v0.12.x

  • Julien: I will have less time to do releases; it needs to become more of a team effort.
  • Alexis: in the long term this will be a responsibility of the build team.
  • Julien: unsure how responsibilities will be decided. LTS will need to sign off and build will need to produce the release.
  • Jeremiah: the iojs/current releases are already a group effort. It’s just that the “long-term” v0.10/v0.12 releases fall on few individuals now.
  • Julien: it’s a bit too much to handle for one person. Also people are sometimes unavailable or on vacation. Would like to have a group of about four people.
  • Ben: more contributors recently signed up. I think Sam Roberts might be interested.
  • Julien: would like to have a release management team.
  • Chris: iojs has had the release manager propose other release managers. Open an issue for this.
  • Resolved as such.

lts: LTS Proposal https://github.com/nodejs/LTS#proposed-lts)/ Proposal: Release Process #1997

  • James: when are we cutting over to the converged stream? Thinking of late august, first LTS release in October. Is this a good time? Most users won’t start migrating until next year because of the holidays.
  • Julien: what are other projects doing, when do they release?
  • James: looking it into it, some do it in fall. No clear pattern.
  • Alexis: what is the benefit of being on a fixed release schedule?
  • James: benefit is it makes planning easier.
  • Trevor: coming from the enterprise side, not having a predictable release schedule isn’t useful.
  • Steven: ICU and Unicode has announced that there will be a yearly release. It’s been helpful for planning.
  • James: It also ties into our regular release schedule and merging next into master etc. The next-to-master merge defines when we can do an LTS release. This should happen at least twice a year. The LTS is cut just before a merge (major bump), so by the time a LTS is cut it should have been stable for half a year.
  • James: please kick tires on this proposal, get feedback from the user communities you’re connected to wrt the frequency and release date.
  • Rod: the TSC should consider the timeframe, and the requirement that there should be two next-to-master merge yearly.
  • Trevor: how does this fit with a 6-week release schedule on master?
  • James: depends on the schedule.
  • Domenic: I don’t see the problem. Just take a 6 months old release and turn it into an LTS.
  • Rod/James/Trevor: because version numbers. The LTS version number needs to be a continuation of a release version.
  • Rod: fixed date, or part of the month.
  • Chris, Rod: get feedback, comment on the issue

Next Meeting

July 15th 2015