io.js TC Meeting 2015-02-04

Agenda

assert: don't compare object prototype property #636 and assert: introduce deepStrictEqual #639 / @vkurchatkin
Release PGP key strategy and policy #709
If time: vm bugs? #548, joyent/node#9084

Bert: Lots of commenting, still working on ICU, investigating binary size and memory size minimisation, playing with transforming the data to C
Chris: Integrating ESLint, held first Streams WG meeting
Colin: Issues and codez, removed all util.is type functions
Isaac: “Real job” stuff
Domenic: Streams WG
Mikeal: Starting Tracing WG, lots of org stuff, roadmap work, early feedback from reaching out to companies is that there is concern about version number conflict between io.js and Node, and debugging/tracing.
Trevor: moving house, tracing
Rod: build collaboration stuff

governance: Add new Collaborators #234 / feedback and more contribs / @chrisdickinson
Stabilization and Release Cycles and Process #405 / further discussion / @iojs/tc

doc: add releases document detail release cycle #630 / proposal from @chrisdickinson

dgram: implicit binds should be exclusive #325 / minor version bump / @bnoordhuis
buffer: implement iterable interface #525 / minor version bump / @bnoordhuis
replace util.isType() with typeof #607 / general use of util.is*() in core re perf
docs: lower the maximum stability level to "stable" (https://github.com/nodejs/io.js/pull/633)
maintain our own package registries for io.js related packages (https://github.com/nodejs/io.js/issues/640#issuecomment-71882645)
Working Groups PR (https://github.com/nodejs/io.js/pull/599)
Remove “unstable” from messaging (https://github.com/nodejs/website/issues/108)
Node.js and io.js should not break each other (no, they shouldn't)
Working group reports

the future of assert module;
relationship between assert and CommonJS Unit Testing spec;
particular #636 bug;
what "deep equality" means and how it should deal with prototypes;
#639 proposal
Isaac: CommonJS is gone & dead, good ideas have to stand on their own
Mikeal: assert probably shouldn’t be in core
Isaac: it’s useful for testing Node itself and useful for runtime asserts
Rod asked if anyone had any objections to the changes because perhaps Validmir should just be given a blessing to go ahead and improve it.
Domenic: medium risk, low reward, perhaps slate this for a 2.0 release.
Lots of discussion about the technical merits
Ben: the a.prototype !== b.prototype is clearly a bug in the initial implementation, it should have been checking __proto__
Rod: new deepStrictEqual() does that check properly

Conclusion: accept both of

(Contrast #636 with #621, which is a more ambitious change and probably not a good idea.)

Rod summarised how the current release key strategy works and is documented on the README. Asked for feedback on:
- Which keyserver(s) to use
- Whether to go with a master/sub key design for the project
No strong opinions, nothing actionable here

Domenic introduced #548, some odd VM bugs that seem to be V8 bugs,
Isaac encouraged Domenic to follow this up with the V8 team to get feedback on how this works