thakis added inline comments.
================
Comment at: test/CodeGen/pch-dllexport.cpp:55
+template <typename T> void __declspec(dllexport)
explicitInstantiationDefAfterDecl(T) {}
+extern template void explicitInstantiationDefAfterDecl<int>(int);
+
----------------
hans wrote:
> thakis wrote:
> > This has two interesting cases:
> >
> > 1. An explicit instantiation declaration.
> > 2. An explicit instantiation definition.
> >
> > You have a test for 2, but not for 1, as far as I can tell (?). For 1, the
> > inline function from the pch-obj should be used (which probably already
> > works; if not a FIXME is enough).
> It's really testing the "explicit instantiation def after explicit
> instantiation decl" case.
>
> With just an explicit instantiation decl (1), there's not much to test, since
> that's like a promise that someone else is going to instantiate the template.
> Nothing needs to be emitted for that.
>
> Maybe I'm not understanding your question?
> Nothing needs to be emitted for that.
Right, I meant including a test that we really don't emit anything for that.
But feel free to omit it if you don't think it adds much. I suggested it
because we end up with TSK_ExplicitInstantiationDeclaration in that case, and
you're checking the TSK in this change, and you have coverage for the other
kinds.
https://reviews.llvm.org/D48426
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
