================
@@ -87,6 +85,28 @@ bool isStdVariant(const Type *Type) {
return isStdType(Type, llvm::StringLiteral("variant"));
}
+bool isStdAny(const Type *Type) {
+ return isStdType(Type, llvm::StringLiteral("any"));
+}
+
+bool isVowel(char a) {
+ switch (a) {
+ case 'a':
+ case 'e':
+ case 'i':
+ case 'o':
+ case 'u':
+ return true;
+ default:
+ return false;
+ }
+}
+
+llvm::StringRef indefiniteArticleBasedOnVowel(char a) {
+ if (isVowel(a))
+ return "an";
+ return "a";
+}
----------------
DonatNagyE wrote:
I think parenthesized `a(n)` in 100% of the messages is uglier than incorrectly
applied `a` or `an` in <1% of the messages (in average English text, 1% of the
words start with `u` [1] -- this may be different for type names, but the order
of magnitude should be similar).
However, this is just bikesheddig, so I won't block any of the three solutions
(decision based on `isVowel()`, using `a(n)` and using filler words to dodge
the issue).
[1] https://www.johndcook.com/blog/2021/08/16/initial-letter-frequency/
https://github.com/llvm/llvm-project/pull/76580
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
