Unexpected panic in subtyping... about flux HOT 4 CLOSED

The reason is that the function expects a strong reference but you only have a mutable one. This (again) can be solved by implementing opening of mutable references (seeing a mutable reference as strong during the duration of the lifetime). Maybe I should add implementing this to my new year's resolutions.

Note that the signature fn(&mut FdMap[@dummy], k: usize) is wrong because the function doesn't preserve the size of the vector, thus you could avoid the panic by taking a strong reference and exposing the updated length in the ensures clause. Then, whether to take strong vs mutable references (provided that we implement opening of mutable references) is a matter of encapsulation, you either expose the length or not.

from flux.

from flux.

Yes, I always wanted to implement it so I never bother making it a user facing error.

from flux.

Ah, ok sounds good let me close. The thing that motivated this issue works, but the original problem still remains, namely:

ctx.fdmap.delete(...)

causes a

panicked at 'expected type at Some(src/wrappers.rs:98:5: 98:27 (#0))', flux-typeck/src/type_env/paths_tree.rs:568:23

which requires me to do make the delete a method of ctx and do ctx.fdmap_delete... [update having a hard time replicating the problem in a small standalone test so will tell you about it offline...]

from flux.