Remove TraitStore from ty_trait #14869
Conversation
|
Interesting test failure... |
| let bounds = parse_bounds(st, |x,y| conv(x,y)); | ||
| assert_eq!(next(st), ']'); | ||
| return ty::mk_trait(st.tcx, def, substs, store, bounds.builtin_bounds); | ||
| return ty::mk_trait(st.tcx, def, substs, bounds.builtin_bounds); |
There was a problem hiding this comment.
Two spaces between substs and bounds.builtin_bounds. This might show up in other places, too.
|
r? anyone? |
|
Wut? That doesn't happen locally :-( |
|
/me will read |
| // FIXME(12470) If we have mutable references to trait objects, we | ||
| // used to use covariant subtyping. I have preserved this behaviour, | ||
| // even though it is probably incorrect. So don't go down the usual | ||
| // path which would require invariance. |
There was a problem hiding this comment.
I've been thinking about this lately, as we briefly discussed on the phone. On the one hand, there are rules against assigning lvalues of unsigned type, so in fact an &mut U(where U is unsized) is more limited than an &mut S (where S is sized). In particular, you cannot do *x = foo as Trait or something like that. For this reason, I believe that e.g. &mut (Trait1+Trait2) <: &mut Trait1 is sound, though I want to consider it more deeply. Nonetheless, I am not happy about the rules for &mut T considering whether T is sized.
I am not convinced this is related to #12470 -- can you open a separate bug perhaps to settle variance w/r/t DST and &mut more definitively?
Also, I presume that when you didn't have this code, various things broke?
There was a problem hiding this comment.
Yeah, at first I thought this was sound too. I couldn't come up with a counter-example. So now I'm not sure.
#12470 is about enforcing invariance here for lifetimes, whereas this is about enforcing invariance for bounds. I'll open a new bug for issue.
Yep, stuff broke. Not sure how much.
Use ty_rptr/ty_uniq(ty_trait) rather than TraitStore to represent trait types. Also addresses (but doesn't close) rust-lang#12470. Part of the work towards DST (rust-lang#12938). [breaking-change] lifetime parameters in `&mut trait` are now invariant. They used to be contravariant.
The rust-lang#14869 removed `TraitStore` from `ty_trait` and represented trait reference as regular `ty_rptr`. An old bug of the missing constraint upon lifetime parameter of trait reference then is fixed as a side effect. Adds tests for affected bugs and closes them. Closes rust-lang#12470. Closes rust-lang#14285.
The rust-lang#14869 removed `TraitStore` from `ty_trait` and represented trait reference as regular `ty_rptr`. An old bug of the missing constraint upon lifetime parameter of trait reference then is fixed as a side effect. Adds tests for affected bugs and closes them. Closes rust-lang#12470. Closes rust-lang#14285.
Use ty_rptr/ty_uniq(ty_trait) rather than TraitStore to represent trait types.
Also addresses (but doesn't close) #12470.
Part of the work towards DST (#12938).