[llvm-branch-commits] [clang] [Clang] [C++26] Expansion Statements (Part 6: Destructuring Expansion Statements) (PR #169685)

[Shafik Yaghmour via llvm-branch-commits]

================
@@ -234,6 +234,46 @@ static IterableExpansionStmtData TryBuildIterableExpansionStmtInitializer(
 #endif
 }
 
+static StmtResult BuildDestructuringDecompositionDecl(
+    Sema &S, Expr *ExpansionInitializer, SourceLocation ColonLoc,
+    bool VarIsConstexpr,
+    ArrayRef<MaterializeTemporaryExpr *> LifetimeExtendTemps) {
+  auto Ctx = Sema::ExpressionEvaluationContext::PotentiallyEvaluated;
+  if (VarIsConstexpr)
+    Ctx = Sema::ExpressionEvaluationContext::ImmediateFunctionContext;
+  EnterExpressionEvaluationContext ExprEvalCtx(S, Ctx);
+
+  // The declarations should be attached to the parent decl context.
+  Sema::ContextRAII CtxGuard(S, S.CurContext->getParent(),
+                             /*NewThis=*/false);
+
+  UnsignedOrNone Arity =
+      S.GetDecompositionElementCount(ExpansionInitializer->getType(), ColonLoc);
+
+  if (!Arity)
+    return StmtError();
+
+  QualType AutoRRef = S.Context.getAutoRRefDeductType();
+  SmallVector<BindingDecl *> Bindings;
+  for (unsigned I = 0; I < *Arity; ++I)
+    Bindings.push_back(BindingDecl::Create(
+        S.Context, S.CurContext, ColonLoc,
+        S.getPreprocessor().getIdentifierInfo("__u" + std::to_string(I)),
----------------
shafik wrote:

So naming these things has me asking myself some questions. I think naming them does make sense but what happens if we have a local inside a expansion and it has the same name will they conflict? 

How does the line numbering look for a diagnostic?

I would like to see tests of this scenario with a local inside the expansion with a name matching this. Yes, it would be IFNDR but we should test it.

https://github.com/llvm/llvm-project/pull/169685