WIP implement a machine readable print-type-info API

[Gankra]

This is very messy, with some commented out stuff and some TODOs. I'm posting this now to get feedback on the approach, and so someone can tell me where stuff should go or if I'm doing some things fundamentally wrong. Note that I'm currently clobbering -Zprint-type-sizes to avoid doing driver plumbing atm.

Background Proposal (doesn't completely match impl, see below)

Draft Documentation

Sample output on gecko/gfx/webrender_bindings

Differences from background proposal:

• Added Opaque, PrimitiveInt, and PrimtiveFloat kinds
• Removed support for "General" enums, as binding to them in a stable way will be hard (can be done in later versions without breaking, per draft docs)
• Backpedaled on "squashing" Rust concepts into C concepts; consumer is expected to do this themselves (see "builtin syntax" in docs)
  • Consequently suppressed more forms from being emitted (see "ommitted builtins" in docs)
• Expected to emit every path a type can be found under (e.g. core::option::Option and std::option::Option)

TODO

• get Visibility of types (help)
• get type/path aliases (help!)
• emit whether a given type is "transparent" for ABI purposes? (Box == *mut)
• set this up as a "target format" flag, instead of -Zprint-type-sizes
• capture public types that aren't used anywhere in the code?
• ensure primtives are always output? (clike enums may be emitted with otherwise unused primitive base-types)

[rust-highfive]

r? @arielb1

(rust_highfive has picked a reviewer for you, use r? to override)

[Gankra]

cc @eddyb @eqrion

[bors]

☔ The latest upstream changes (presumably #42932) made this pull request unmergeable. Please resolve the merge conflicts.

[arielb1]

What's the status of this PR? Do you need any help (I won't be available until this Monday)?

[Gankra]

Hey @arielb1 sorry I pivoted off this a bit and then went on vacation. Tomorrow I get back to work and, unless I get some information stating otherwise, this should be my top priority.

[arielb1]

Hi @gankro are you still working on this? Just a ping to make sure it isn't getting lost.

[carols10cents]

Hiii Ganksy, triage procedure says a PR with >14 days of no author activity gets closed as inactive. Feel free to reopen when you have a chance to get back to this!

[Gankra]

Yeah I'll get back to this once I've got all the fires in webrender put out!

[eddyb]

I have a branch in progress which if I get out soon would provide a much better baseline for this.

[Gankra]

@eddyb's branch became #45225, for reference.

[Gankra]

(continuing this work is blocked on that PR)

[RReverser]

@gankro Given that the referenced PR was merged couple of months ago, are there any other blockers for this one left? (apart from free time, of course 😄 )

[Gankra]

My enum RFC combined with cbindgen relieves a lot of pressure on this, and i'm recovering on indeterminate-length leave right now, so if you're interested I suggest taking this over.