[PATCH] D61837: Make it possible control matcher traversal kind with ASTContext

[Aaron Ballman via Phabricator via cfe-commits]

aaron.ballman added inline comments.


================
Comment at: include/clang/ASTMatchers/ASTMatchers.h:718
+template <typename T>
+internal::Matcher<T> traverse(ast_type_traits::TraversalKind TK,
+                              const internal::Matcher<T> &InnerMatcher) {
----------------
steveire wrote:
> aaron.ballman wrote:
> > steveire wrote:
> > > aaron.ballman wrote:
> > > > Is this an API we should be exposing to clang-query as well? Will those users be able to use a string literal for the traversal kind, like they already do for attribute kinds (etc)?
> > > Yes, I thought about that, but in a follow-up patch. First, I aim to extend the `TraversalKind` enum with `TK_IgnoreInvisble`.
> > This is new functionality, so why do you want to wait for a follow-up patch (is it somehow more involved)? We typically add support for dynamic matchers at the same time we add support for the static matchers because otherwise the two get frustratingly out of sync.
> Perhaps I can add the support in DynamicASTMatchers. Can you give some direction on that? I don't recall where the attr strings are handled.
> 
> A corresponding change to `clang-query` will be in another patch anyway because it's a different repo.
> Perhaps I can add the support in DynamicASTMatchers. Can you give some direction on that? I don't recall where the attr strings are handled.

I always wind up having to use svn blame to figure it out, so you're not alone in your recollections. ;-) You add a new specialization of `ArgTypeTraits` to Marshallers.h for the enumeration type. See around line 123 or so for how it's done for `attr::Kind`.

> A corresponding change to clang-query will be in another patch anyway because it's a different repo.

Once you have the enumeration being marshaled, I believe all you have to do is expose the new API via Registry.cpp like any other matcher, though I'd also request tests be added to Dynamic/RegistryTest.cpp since this is adding a small amount of new marshaling code.


Repository:
  rC Clang

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D61837/new/

https://reviews.llvm.org/D61837