haraken@, would you agree that this sounds like something that could be implemented in blink-gc clang plugin?  I think this route sounds promising - if we can implement the equivalent of static_asserts from wtf/PtrUtil.h in blink-gc clang plugin, then wtf/PtrUtil.h can have no function bodies and can just say:

    namespace WTF {
    
    template <typename T>
    using makeUnique = base::MakeUnique<T>
    
    template <typename T>
    using wrapUnique = base::WrapUnique<T>
    
    ...
    
    }

IMO, the static asserts cannot be moved into //base (because //base doesn't know anything about things like WTF::IsGarbageCollectedType<T>.

There's a cost to implementing checks in clang: it doesn't work in MSVC, it's harder to make changes, any changes take longer to propagate through the system, and it can make compiler rolls harder.

It is a bit unfortunate to have this bit of duplication, but it doesn't feel overly burdensome either.

https://codereview.chromium.org/2547053003/#msg16
haraken@ wrote: 
> What's the advantage of removing WTF::wrapUnique, in the first place? We are not
> allowed to use base:: types directly in core/ and modules/, so we anyway need to
> wrap the base:: types in wtf/.

Right.  There would be no much benefit because we limit base:: usage in Blink.  Just code duplication is harmful in general.
So if we need much effort to consolidate implementations into one, I don't think we need to do so.

Ok, blink-gc plugin isn't reasonable.
I think it's not difficult to move IsGarbageCollectedType to base.

wrapUnique is so small and obvious that I don't care about code duplication.
I'm in favor of WONTFIX-ing this.

Also, Clang plugins should be used for checks that cannot be expressed in C++.
If the equivalent check can be written in C++, the C++ version would be superior
in all respects.

Related: https://bugs.chromium.org/p/chromium/issues/detail?id=781645