[PATCH] D114562: [clang][docs] Inclusive language: remove use of sanity check in option description

[Aaron Ballman via Phabricator via cfe-commits]

aaron.ballman accepted this revision.
aaron.ballman added a comment.
This revision is now accepted and ready to land.

LGTM!



================
Comment at: clang/docs/ThreadSafetyAnalysis.rst:471
 
-  + ``-Wthread-safety-attributes``: Sanity checks on attribute syntax.
+  + ``-Wthread-safety-attributes``: Validation checks on attribute syntax.
   + ``-Wthread-safety-analysis``: The core analysis.
----------------
aaronpuchert wrote:
> Quuxplusone wrote:
> > I'd just say "Rejects invalid attribute syntax." But also
> > - surely this is totally obvious, that invalid syntax causes an error? why do we even need to write this?
> > - line 469 says "three" but then there are four bullet points
> It's more about semantics than syntax, i.e. something like “this attribute doesn't make sense on such a declaration.” The full (current) list of warning messages is on https://clang.llvm.org/docs/DiagnosticsReference.html#wthread-safety-attributes. (Even “ignoring //A// attribute because its argument is invalid” is actually just saying that we don't understand the AST.)
> 
> "Validation checks" feels like a hendiadys and doesn't really add anything. Indeed something like “Semantic (checks|validation) for thread safety attributes” would be most accurate. Though maybe @aaron.ballman has a better idea.
TIL what `hendiadys` means.

I think the currently proposed wording is reasonable (it's not just about syntax, as @aaronpuchert says, it's also about semantic checks). Really, this groups is about diagnostics that relate to incorrect use of the attribute itself, not the analysis supported by the attribute and "semantic checks" works for that).


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D114562/new/

https://reviews.llvm.org/D114562