reflect: TypeOf(unsafe.Pointer(nil)).PkgPath() returns "", not "unsafe"

[mdempsky]

The documentation for reflect.Type.PkgPath says:

    // PkgPath returns a defined type's package path, that is, the import path
    // that uniquely identifies the package, such as "encoding/base64".
    // If the type was predeclared (string, error) or not defined (*T, struct{},
    // []int, or A where A is an alias for a non-defined type), the package path
    // will be the empty string.

unsafe.Pointer is not listed as a predeclared type: https://golang.org/ref/spec#Predeclared_identifiers (Though go/types lists it as one: https://golang.org/pkg/go/types/#BasicKind)

So it seems to reason that PkgPath should return "unsafe", but it instead returns "": https://play.golang.org/p/ZnUOgHjwz58

/cc @griesemer @ianlancetaylor

[odeke-em]

@mdempsky, thank you for catching this discrepancy, nice work! If I may ask, perhaps should we repurpose this as a proposal to include unsafe.Pointer into builtin but also to update the spec? I think this was just an oversight from commit 20850fc almost 12 years ago in which we perhaps forgot to update the spec. @robpike @rsc @griesemer @ianlancetaylor is my assumption correct? Thank you.

[mdempsky]

unsafe.Pointer is mentioned in the spec. It's also declared in package unsafe, not the universe, so it's properly not listed as a predeclared type/identifier. I don't see any reason to change the spec here.

[ianlancetaylor]

At first glance this seems like a compiler bug to me. It's probably treating unsafe.Pointer as a predeclared type rather than as a type defined in the "unsafe" package, so it's probably generating an uncommonType struct with a zero pkgPath field. Or something like that.

[mdempsky]

I can believe it's actually a compiler bug.

FWIW, it looks like with gccgo 11, both PkgPath and also Name return "".

[ianlancetaylor]

For gccgo it's a libgo bug, because libgo holds the unique type descriptor for unsafe.Pointer (https://go.googlesource.com/gofrontend/+/refs/heads/master/libgo/runtime/go-unsafe-pointer.c). But as far as I know the gc compiler doesn't work that way.

[mdempsky]

Sure. I'm just pointing out a similar issue also affects gccgo.

[zigo101]

It looks Go spec views unsafe things as builtin things in several places.

• ... or the result value of some built-in functions such as unsafe.Sizeof applied to any value, ...
• The following built-in functions are not permitted in statement context: ... unsafe.Alignof unsafe.Offsetof unsafe.Sizeof

[gopherbot]

Change https://golang.org/cl/313349 mentions this issue: cmd/compile: fix wrong package path for unsafe.Pointer

[gopherbot]

Change https://golang.org/cl/313890 mentions this issue: all: clarify that unsafe.Pointer is not a predeclared type