[WHLSL] The checker should not have pointers/references into its maps

[Robin Morisset]

m_typeMap is currently HashMap<AST::Expression*, ResolvingType> m_typeMap;
This is buggy, because some code in the checker has references into it, and they get invalidated whenever the typeMap is resized. Instead the Map should contain UniqueRef<ResolvingType>, and then we should only have references to the ResolvingType itself.
This bugs currently prevents checking the full standard library.

[Robin Morisset]

Created attachment 370812 [details]
WIP

[Myles C. Maxfield]

Comment on attachment 370812 [details]
WIP

View in context: https://bugs.webkit.org/attachment.cgi?id=370812&action=review

> Source/WebCore/Modules/webgpu/WHLSL/WHLSLChecker.cpp:494
> +    HashMap<AST::Expression*, std::unique_ptr<ResolvingType>> m_typeMap;

UniqueRef didn't work?

[Robin Morisset]

No, because it does not offer the right methods to be in a HashMap (I think it is missing a copy operator or something. I could have looked in detail, and tried to improve either HashMap or UniqueRef to work together, but using std::unique_ptr seemed way easier and good enough for this purpose.

(In reply to Myles C. Maxfield from comment #2)
> Comment on attachment 370812 [details]
> WIP
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=370812&action=review
> 
> > Source/WebCore/Modules/webgpu/WHLSL/WHLSLChecker.cpp:494
> > +    HashMap<AST::Expression*, std::unique_ptr<ResolvingType>> m_typeMap;
> 
> UniqueRef didn't work?

[Radar WebKit Bug Importer]

<rdar://problem/51288768>

[Saam Barati]

patch forthcoming with tests.

[Saam Barati]

actually, just gonna do this as part of the matrix patch. it's too difficult to split up as we'll end up with checker errors.

[Saam Barati]

*** This bug has been marked as a duplicate of bug 198876 ***