LegalizeAdulthood added inline comments.
================
Comment at: unittests/ASTMatchers/ASTMatchersTest.cpp:4990
@@ +4989,3 @@
+ EXPECT_TRUE(matches("typedef int hasUnderlyingTypeTest;",
+ typedefDecl(hasUnderlyingType(asString("int")))));
+ EXPECT_TRUE(
----------------
aaron.ballman wrote:
> I would expect hasUnderlyingType to look through all of the type sugar, not
> just the top layer of it (since existing matchers can implement the current
> behavior by using hasType, I believe). I think the correct approach is to get
> the underlying type, then loop to see whether that type matches, and if not,
> strip off a layer of sugar and try again. Terminate the loop when there's no
> more sugar to strip. The following should all match:
> ```
> EXPECT_TRUE(
> matches("typedef int foo; typedef foo bar; typedef bar baz;",
> typedefDecl(hasUnderlyingType(asString("int")),
> hasName("bar"))));
> EXPECT_TRUE(
> matches("typedef int foo; typedef foo bar; typedef bar baz;",
> typedefDecl(hasUnderlyingType(asString("foo")),
> hasName("baz"))));
> EXPECT_TRUE(
> matches("typedef int foo; typedef foo bar; typedef bar baz;",
> typedefDecl(hasUnderlyingType(asString("int")),
> hasName("baz"))));
> ```
> The other question I have is about qualifiers. Should this match?
> ```
> EXPECT_TRUE(
> matches("typedef const int foo;",
> typedefDecl(hasUnderlyingType(asString("int")),
> hasName("foo"))));
> ```
> Or should it require `asString("const int")`?
> (Regardless of the answer, we should document the behavior and add a test
> with qualifiers.)
I'm happy to make the improvement, but I don't know how. I simply call the
`Node.underlyingType()`, just like `hasType` calls `Node.getType()`. I don't
know why they are different.
http://reviews.llvm.org/D8149
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
