aaron.ballman added inline comments.

================
Comment at: clang/include/clang/Basic/AttrDocs.td:6473-6475
+The HLSL ``resource`` attribute is not valid to manually specify in
+HLSL. It is applied by the compiler to HLSL resource type objects enabling them
+to be handled appropriately in CodeGen.
----------------
beanz wrote:
> aaron.ballman wrote:
> > Because the attribute cannot be written by the user, it seems like we 
> > shouldn't bother documenting this attribute at all. WDYT? If you agree, you 
> > can mark the attribute as `Undocumented` in Attr.td.
> Yea... I am kinda split mind on this myself, and I think the tl;dr is that 
> this is the wrong place to document this.
> 
> The existing HLSL compiler and HLSL language are woefully short on 
> documentation, so I'm trying to at least make a best-effort at documenting 
> things as I implement them in clang. Since the AttrDocs are really intended 
> to be user-facing documentation this is probably the wrong place to document 
> it.
> 
> I'll pull this out and maybe separately create a folder under clang's docs 
> directory for HLSL-related documentation so that it lives with the more 
> project-centric docs. Does that make sense?
> I'll pull this out and maybe separately create a folder under clang's docs 
> directory for HLSL-related documentation so that it lives with the more 
> project-centric docs. Does that make sense?

I'm fine with that approach. FWIW, I've struggled with this a bit as well. We 
want all new attributes to be documented, and it's a sometimes super fuzzy line 
between "this is an implementation detail" and "users can use this with wild 
abandon". However, in this case, because the attribute has no spelling, it's 
pretty clearly an implementation detail. So putting that into the "using clang 
as a library" or other internal documentation.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D130018/new/

https://reviews.llvm.org/D130018

_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to