[PATCH] D103789: [clangd] Type hints for C++14 return type deduction

[Nathan Ridge via Phabricator via cfe-commits]

nridge added inline comments.


================
Comment at: clang-tools-extra/clangd/InlayHints.cpp:79
+      if (!Deduced.isNull()) {
+        SourceRange R = D->getReturnTypeSourceRange();
+        // For operator auto(), have to get location of `auto` a different way.
----------------
sammccall wrote:
> nit: bool TrailingReturnType = D->getReturnTypeSourceRange().isValid()?
Sorry, I don't really understand this comment.

If you're thinking about the test case where a trailing return type is present and we avoid producing a hint in that case (`f3` in `TypeHints.ReturnTypeDeduction`), that happens because `getContainedAutoType()` returns null, so we don't get into this block at all.


================
Comment at: clang-tools-extra/clangd/unittests/InlayHintTests.cpp:505
+  )cpp",
+      ExpectedHint{": int", "ret1a"}, ExpectedHint{": int", "ret1b"},
+      ExpectedHint{": int &", "ret2"}, ExpectedHint{": int", "retConv"});
----------------
sammccall wrote:
> This reads as "auto[: int] f1(int x);", which doesn't look much like familiar syntax, C++ or otherwise.
> 
> I guess we could try `auto f1(int x)[-> int];`?
> 
> (From playing with these in vscode, I'm not sure I find the punctuation very useful)
I like this suggestion, thanks!

I agree it's nice if the hint is in a place where it would have been valid syntax.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103789/new/

https://reviews.llvm.org/D103789