================
@@ -3703,8 +3703,14 @@ void Parser::ParseDeclarationSpecifiers(
// We reject AT_LifetimeBound and AT_AnyX86NoCfCheck, even though
they
// are type attributes, because we historically haven't allowed these
// to be used as type attributes in C++11 / C23 syntax.
- if (PA.isTypeAttr() && PA.getKind() != ParsedAttr::AT_LifetimeBound
&&
- PA.getKind() != ParsedAttr::AT_AnyX86NoCfCheck)
+ if (PA.getKind() == ParsedAttr::AT_LifetimeBound) {
+ Diag(PA.getLoc(), diag::err_attribute_wrong_decl_type_str)
+ << PA << PA.isRegularKeywordAttribute()
+ << "parameters and implicit object parameters";
----------------
ilya-biryukov wrote:
I would like to second @cor3ntin's comment. We should aim to have all message
strings in one place (i.e. the .td file), it makes searching for error messages
much easier.
However, `err_attribute_wrong_decl_type_str` in particular seems to be using
strings in the source code
([1](https://github.com/llvm/llvm-project/blob/0cda970ecc8a885acf7298a61370a1368b0ea39b/clang/lib/Sema/SemaDeclAttr.cpp#L5529),
[2](https://github.com/llvm/llvm-project/blob/0cda970ecc8a885acf7298a61370a1368b0ea39b/clang/lib/Sema/SemaDeclAttr.cpp#L1842),
[3](https://github.com/llvm/llvm-project/blob/0cda970ecc8a885acf7298a61370a1368b0ea39b/clang/lib/Sema/SemaDeclAttr.cpp#L2736)).
So I think this change is consistent with that, and I think we should just land
it and replace this with a follow-up instead (@emaxx-google would you be
willing to do it?)
Adding another `AttributeDeclKind` seems off, given that it's a enumeration
used primarily by parser. But adding a new enumeration for select (that maps to
`AttributeDeclKind` for the known values and adds a few more) seems like a good
path forward.
https://github.com/llvm/llvm-project/pull/118567
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
