Details
- Reviewers
aaron.ballman
Diff Detail
- Repository
- rG LLVM Github Monorepo
- Build Status
Buildable 188246 Build 284407: arc lint + arc unit
Event Timeline
Can you split this out into two separate reviews? That makes it easier for the reviewers so we don't mix up context as well as when we do code archeology for changes. (I also made some quick comments as I noticed things in the file, but I've not done a full review yet.)
clang/include/clang/AST/Decl.h | ||
---|---|---|
2961 | ||
clang/include/clang/Basic/Attr.td | ||
3490 | Type attribute? That seems surprising -- this is probably meant to be InheritableAttr? | |
3497 | Hmmm, very odd that the msvc no unique address attribute is only supported on Itanium ABI targets. Should this be TargetMicrosoftCXXABI instead? | |
clang/include/clang/Basic/DiagnosticSemaKinds.td | ||
3366–3368 ↗ | (On Diff #462234) | You shouldn't need to add this diagnostic -- if the Subjects list is correct in Attr.td, we automatically diagnose this sort of thing. |
Sure, done: https://reviews.llvm.org/D134475
clang/include/clang/Basic/Attr.td | ||
---|---|---|
3497 | My fault, I did blind copy-paste completely forgetting about ABI! |
clang/include/clang/Basic/Attr.td | ||
---|---|---|
3492 | Does the 201803 value match what MSVC returns from __has_cpp_attribute? | |
clang/test/AST/msvc-attrs.cpp | ||
4 | Thanks for the start to the tests! You should also add CodeGen tests as well as Sema tests. The CodeGen tests are to make sure we're codegenning the correct offsets to members, and the Sema tests are to ensure we're matching diagnostic behavior with MSVC rather than accepting code MSVC rejects. |
Abandoning due to lack of time and expertise. I might come back and tackle it in several months.