[NFC][Clang] Adopt TrailingObjects convenience API in MacroArgs

[jurahul]

Adopt convenience API for single trailing type added in #138970.

[llvmbot]

@llvm/pr-subscribers-clang

Author: Rahul Joshi (jurahul)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/139635.diff

1 Files Affected:

• (modified) clang/lib/Lex/MacroArgs.cpp (+2-2)

diff --git a/clang/lib/Lex/MacroArgs.cpp b/clang/lib/Lex/MacroArgs.cpp
index 2f97d9e02bc11..548df16c59f6b 100644
--- a/clang/lib/Lex/MacroArgs.cpp
+++ b/clang/lib/Lex/MacroArgs.cpp
@@ -66,7 +66,7 @@ MacroArgs *MacroArgs::create(const MacroInfo *MI,
                   "uninitialized array (as opposed to reusing a cached "
                   "MacroArgs)");
     std::copy(UnexpArgTokens.begin(), UnexpArgTokens.end(),
-              Result->getTrailingObjects<Token>());
+              Result->getTrailingObjects());
   }
 
   return Result;
@@ -119,7 +119,7 @@ const Token *MacroArgs::getUnexpArgument(unsigned Arg) const {
   assert(Arg < getNumMacroArguments() && "Invalid arg #");
   // The unexpanded argument tokens start immediately after the MacroArgs object
   // in memory.
-  const Token *Start = getTrailingObjects<Token>();
+  const Token *Start = getTrailingObjects();
   const Token *Result = Start;
 
   // Scan to find Arg.

[shafik]

Can you please justify this change, we don't use getTrailingObjects like this anywhere else and I am not sure I see the point in this change. This should always go in the summary so that a reviewer does not have to guess you intent.

In general we prefer being as explicit as possible since that leads to less ambiguity for future developers working on the code later.

[jurahul]

I filled in the description now. I am doing similar changes on LLVM/MLIR side (change has been approved).

I added these convenience APIs to TrailingObjects to support the common code patterns of a single trailing type as well as creating an ArrayRef from the trailing objects pointer.

[jurahul]

FYI, LLVM and MLIR change of similar nature: #138554

[kazutakahirata]

@jurahul Thanks for the pointer in the description. So, in this particular case, we end up invoking the following?

  FirstTrailingType *getTrailingObjects() {
    static_assert(sizeof...(TrailingTys) == 1,
                  "Can use non-templated getTrailingObjects() only when there "
                  "is a single trailing type");
    return getTrailingObjects<FirstTrailingType>();
  }

[jurahul]

@jurahul Thanks for the pointer in the description. So, in this particular case, we end up invoking the following?

  FirstTrailingType *getTrailingObjects() {
    static_assert(sizeof...(TrailingTys) == 1,
                  "Can use non-templated getTrailingObjects() only when there "
                  "is a single trailing type");
    return getTrailingObjects<FirstTrailingType>();
  }

Right. With this API , when there is a single trailing type, you can call getTrailingObjects directly without having to redundantly specify that single trailing type again.

[jurahul]

Closed by mistake.

[jurahul]

Note: I tried to do a change similar to LLVM/MLIR where I tried to convert all usage under clang/AST and it became too large, so will split it up into multiple ones.

[kazutakahirata]

LGTM. Thanks!

[shafik]

I filled in the description now. I am doing similar changes on LLVM/MLIR side (change has been approved).

I added these convenience APIs to TrailingObjects to support the common code patterns of a single trailing type as well as creating an ArrayRef from the trailing objects pointer.

Thank you, the change makes a lot more sense with the context.