[PATCH] D121602: [clang][dataflow] Model the behavior of non-standard optional constructors

[Yitzhak Mandelbaum via Phabricator via cfe-commits]

ymandel added inline comments.


================
Comment at: clang/lib/Analysis/FlowSensitive/Models/UncheckedOptionalAccessModel.cpp:110
+    return false;
+  auto TypeName = Type->getAsCXXRecordDecl()->getQualifiedNameAsString();
+  return TypeName == "std::optional" || TypeName == "absl::optional" ||
----------------
This could get expensive. Maybe add a FIXME to optimize this by avoiding `getQualifiedNameAsString`?


================
Comment at: clang/lib/Analysis/FlowSensitive/Models/UncheckedOptionalAccessModel.cpp:264
+          +[](const CXXMemberCallExpr *E, LatticeTransferState &State) {
+            assignOptionalValue(*E->getImplicitObjectArgument(), State,
+                                State.Env.getBoolLiteralValue(true));
----------------
I realize this is consistent with the previous version, but just noticed this issue. By using `assignOptionalValue` for `emplace` (and `reset`), does that break the following case?

```
optional<int> opt;
if (p) opt.emplace(3);
someCode();
if (p) use(*opt);
```


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D121602/new/

https://reviews.llvm.org/D121602