Tracking issue for moving custom-page-sizes to phase 3

[fitzgen]

The entry requirements for phase 3 are

Entry requirements:

• Test suite has been updated to cover the feature in its forked repo.
• The test suite should run against some implementation, though it need not be the reference interpreter.
• Updates on the actual spec document and reference interpreter are NOT yet required (but can happen earlier).

These entry requirements are technically satisfied today:

• this repo has a test suite that fully covers the custom-page-sizes feature, and
• these tests are passing in Wasmtime's implementation.¹

The next big tasks for advancing this proposal are to

• update the spec,
• implement support in the spec interpreter, and
• implement toolchain support in wasm-ld and compilers using it.

These are all items that fall under the work that happens during phase 3:

During this phase, the following proceeds in parallel:

• Engines implement the feature (where applicable).
• The spec document in the forked repo is updated to include the full English prose and formalization.
• The reference interpreter in the forked repo is updated to include a complete implementation of the feature.
• The feature is implemented in toolchains.
• Remaining open questions are resolved.

So I believe the proposal is more than ready to advance to phase 3.

However, before taking the phase advancement to a CG meeting, I wanted to open up an issue for discussion. Does anyone believe there are any extant phase 2 tasks that should be completed before this proposal advances to phase 3? Thanks!

Footnotes

1. I do not know for sure whether the three other implementations run the spec tests or their own custom test suites. ↩

[fitzgen]

FYI: I've opened a PR to add this to the March 25th agenda.

[titzer]

I took a stab at adding the requisite data types to the reference interpreter but quickly realized we should probably merge in the Wasm 3.0 changes (or rebase) before doing so. There have been a number of generalizations of how memory types are represented, e.g. because of memory64.

[fitzgen]

Oh I should close this issue because the proposal has moved to phase 3 now. And I should maybe also open a phase 4 tracking issue.

@titzer I can merge 3.0 tomorrow (is it still in a separate branch? I can figure it out tomorrow if you don't know off the top of your head.)

[titzer]

Oh I should close this issue because the proposal has moved to phase 3 now. And I should maybe also open a phase 4 tracking issue.

@titzer I can merge 3.0 tomorrow (is it still in a separate branch? I can figure it out tomorrow if you don't know off the top of your head.)

Wasm 3.0 is merged into the main branch of the spec now, yep.

[fitzgen]

Wasm 3.0 merge is at #47