================
@@ -2068,7 +2068,8 @@ bool Lexer::LexNumericConstant(Token &Result, const char 
*CurPtr) {
   }
 
   // If we have a digit separator, continue.
-  if (C == '\'' && (LangOpts.CPlusPlus14 || LangOpts.C23)) {
+  if (C == '\'' &&
+      (LangOpts.CPlusPlus14 || LangOpts.C23 || ParsingPreprocessorDirective)) {
----------------
AaronBallman wrote:

There's not a one-to-one mapping of `LangOptions` to frontend of driver 
options, yes. But I think it's a stretch to argue that the standard mode 
tracked by `LangOptions` is not exposed via flags.

I don't think per-feature `LangOptions` is viable. There's digit separators, 
hex float support, binary literal support, different literal suffixes, various 
keywords, etc and that doesn't seem likely to scale well. Also, it makes it 
seem like we support a la carte language features when we don't -- we 
intentionally don't want to let users disable standard features (in general; 
exceptions exist) and so that approach gives the impression contributors need 
to check both language mode and feature availability which is a maintenance 
concern.

I'm still wondering about the question I asked here: 
https://github.com/llvm/llvm-project/pull/93753#issuecomment-2173666602

https://github.com/llvm/llvm-project/pull/95798
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to