You say there would be a lint warning the user, but IMO it should be reversed. Don't transform the clone and provide a lint notifying the user the clone could maybe be removed.

Love the series! But I have a lot of doubt about this last one. It feels like it adds a lot of subtleties to the language.

Especially that unsafe example you gave seems really bad. Unsafe Rust is already very hard, but when you can't even trust anymore that a clone actually happens...

Yeah, intrusive datastructures are definitely on the list!

Funny idea I had just before I fell asleep:

How unsafe can you go?
An exploration of VERY unsafe code, but it's all good.

I just don't know if I can find enough interesting things to talk about.

Hmmm, not sure!
Maybe check this toggle?

How though? I think my counter is on 7000+ days.
Do you never shut your PC down?

An update has to be pending really long before windows auto-reboots

Ok, so there are two coalitions possible I think.

Center: D66 VVD GL-PVDA CDA
Center right: D66 VVD CDA JA21

I hope for the former.

I had hoped we could do without the VVD for once, but alas.

Exit poll in half an hour!

Probably easiest to look at what an existing HAL does, like embassy

Ha lol, yeah there's lots of cursed stuff around!

On cortex-m, if you have a function at address 0x1000, your function pointer must be 0x1001 or your chip will crash (sometimes) :P

Copy types cannot impl drop. So T: Drop is !Copy.
Not exactly what you're asking, but maybe close enough?