Delegate debug implementation for InternedString

[matklad]

Let's make InternedString debug implementation the same as for String / str? It's more concise, which helps when you debug printing other stuff, like eprintln!("id = {:?}", a_package_id).

[rust-highfive]

r? @alexcrichton

(rust_highfive has picked a reviewer for you, use r? to override)

[matklad]

@bors delegate=Eh2406

cc @Eh2406 :)

[bors]

✌️ @Eh2406 can now approve this pull request

[Eh2406]

I originally added the extra verbiage to help debugging cases were InternedString and str act different. But now that they hash the same, I don't know of any place that will get past the type checker.
@bors r+

[bors]

📌 Commit 111024c has been approved by Eh2406

[bors]

⌛ Testing commit 111024c with merge 9895340...

[matklad]

Hm, @Eh2406, could we implement InternedString as pub struct InternedString(&'static str)? That should allow us to get rid of some unsafe blocks here perhaps?

[Eh2406]

Yes it could, and I'd be open to a PR that does.
It was done this way to have things by default be the fast pointer comparison rather than the full string comparison.
But:

• Eq I believe already has a fast path for the pointers being equal.
• Hash we had to back out, apparently cargo requires things to Hash the same across program executions.
• !Eq we know that interned strings have been duplicated, if the pointers differ than the contents differ.

So we would switch from overriding hash to overriding EQ.

[matklad]

It was done this way to have things by default be the fast pointer comparison rather than the full string comparison.

I think this is could be implemented if we store &static str and implement PartialEq manually: https://play.rust-lang.org/?gist=b55374c5e2d04aef28d385f5bee22ec1&version=stable

[bors]

☀️ Test successful - status-appveyor, status-travis
Approved by: Eh2406
Pushing 9895340 to master...

[Eh2406]

Completely agree, that would work.