Robin Lindner

I license past and future contributions under the triple MIT OR Apache-2.0 OR Zlib license, allowing licensees to chose any one at their option.

There are still a few things to do in the associated tracking issue (https://github.com/rust-lang/rust/issues/57241), but guess it won't be long before this becomes stable.

@333fred Can you add a `nuint` overload to your proposal? A `nint` overload is already covered by the existing [`IntPtr` overload](https://docs.microsoft.com/en-us/dotnet/api/system.threading.interlocked.compareexchange?view=net-6.0#system-threading-interlocked-compareexchange(system-intptr@-system-intptr-system-intptr))). But it might be useful to add CompareExchangeNInt (@tannergooding...

Does this API require JIT-changes?

> Could bool and char overloads be also considered? They'd be trivial to implement with this. In my opinion, a good idea. What do you think? @terrajobst @stephentoub

@stephentoub Oh I thought something like that always has to be strictly re-reviewed. I ask because I plan to implement this API if no one objects.

> I'm concerned this may be too complex. What is your plan? You are right. I took a look at the whole thing. I would probably have to include instructions...

If we use a Guid, I think NewGuid is a wrong option. NewGuid tries to generate a cryptographically secure random Guid. Performance-wise this would not be fatal compared to a...

> Name collisions would not be a problem if the name contains a Guid. Also a GUID can collide even if it is only every 340_282_366_920_938_463_463_374_607_431_768_211_455th case.

 same problem