Stage 2.7 reviews

[michaelficarra]

I'm planning on presenting this proposal for Stage 2.7 at the upcoming plenary. See the slides. I would appreciate if at least 2 of you could review the spec text (as well as #14/#15/#16) before then.

• @acutmore
• @ljharb
• @jessealama

[acutmore]

I have reviewed the spec and the 3 PRs. No notes.

[acutmore]

For the slides, it lists that Kotlin and itertools-ts have parameterised modes - might be worth highlighting which one is the default. Checking their docs it looks like Kotlin defaults to 1 and itertools-ts default to 3.

[jessealama]

I took a look at the spec as well as the PRs. They're a sensible approach to dealing with the edge cases. LGTM

[ljharb]

Review done:

• spec: looks like "chunks" will return an undersized chunk when an uneven-split iterator is done, and "windows" will drop the unused values
• throw on .next() when windows cannot yield at least 1 full window #14
• yield an undersized window when windows cannot yield at least 1 full one #15
• pad an undersized first window with undefined #16

As a delegate, I think I've come around to either preferring #16, or better, a secret fourth option: make the done: true iterator result have an array of the remaining values as the value.