Consider adding Windows Debug Interface Access (DIA) SDK wrappers for PDB analysis

[jieyouxu]

My only concern with the pdb crate is it doesn't look like it's been updated in awhile. It's probably fine (it has few dependencies by the looks of it) but it makes me slightly wary.

[..] I think long term it would be good to use official debugging APIs on Windows hosts and only fallback to the pdb crate on other platforms (or maybe if they're unavailable for some reason).

-- #143669 (comment)

Currently, some run-make tests use llvm-pdbutil to inspect PDB for symbols and other purposes, but operating on llvm-pdbutil's textual output seems rather fragile, as between llvm-pdbutil versions that's not guaranteed to stay the same (there's an pdb2yaml subcommand, but then we're still parsing yaml?).

Consider introducing Windows Debug Interface Access (DIA) API wrappers to perform PDB analysis programmatically via the Officially Recommended ™️ way instead.

[Fulgen301]

Microsoft has an official crate for DIA.

[jieyouxu]

Oh cool, I didn't realize that was a thing :D

EDIT: looking at it, it seems still fairly raw/experimental at the moment, but worth keeping an eye on.

[Fulgen301]

it seems still fairly raw/experimental at the moment

Unless I missed a label, it shouldn't be - calling windows-bindgen via a build script is the recommended way to get bindings for a third party winmd, and the crate does just that. And at least the example looks exactly like if it was part of windows instead of its own crate.

[ChrisDenton]

It's "raw" in the sense that it directly wraps DIA apis so you need to use unsafe and figure out safe wrappers yourself. But other that that, yeah, they're decent, well maintained bindings.