@WG-unsafe-code-guidelines what do you think would be a good place to collect assumptions that rustc makes about MIR semantics? And what would that list contain so far?
Looks like #61922 will add another item. And then there is #60187 which however I think just assumes that after a
StorageLive, the address of the local is allowed to differ. (There used to be more assumptions that however were a problem and thus got reverted in #61373).
That's an interesting idea
@Taylor Cramer if you think something is missing from this list, please let us know
It seems like to start we could just use an issue
@RalfJ wasn't somebody interested / working on a MIR model ?
probably several people? don't know what you are referring to
operational semantics for MIR
no I mean what project / which people
but also I think this is not a 1-step task
or a memory model for MIR or similar
I have in my head a description of what I think the interface between the MIR semantics and the memory model should look like
which is important because it fixes a lot of terminology without fixing all the behavior
nono, i was just wondering if something like that was already partially being written somewhere, because maybe that place could also be used to document these things
and at some point I'll have the time to write that down
not that I know of