Implement trans for the MIR Switch terminator #30337
Conversation
|
r? @arielb1 (rust_highfive has picked a reviewer for you, use r? to override) |
| Abc::C => 3, | ||
| Abc::D => 4, | ||
| Abc::B(_) => 2, | ||
| Abc::A(_) => 1, |
There was a problem hiding this comment.
You need at least a test for the _ => case.
|
@nagisa Fixed nit and added test case for @dotdash I think I fixed this in my last commit. Does that look right to you? |
|
@wesleywiser @dotdash regarding
Re EDIT: ah, right, it might be a Thinking about that more, |
|
@nagisa AIUI, the fact that the diverge block can contain an |
|
@dotdash but all you can really do in that block, really, is having a I’d rather have |
|
So, it seems to me that perhaps we should just remove |
|
As far as having a special block for unreachable code, that seems fine, but it seems like something specific to trans, and not necessarily something that needs to be made manifest in the MIR. |
| unimplemented!() | ||
| mir::Terminator::Switch { ref discr, ref adt_def, ref targets } => { | ||
| let adt_ty = bcx.tcx().lookup_item_type(adt_def.did).ty; | ||
| let represented_ty = adt::represent_type(bcx.ccx(), adt_ty).clone(); |
There was a problem hiding this comment.
Nit: I don't think you need to call clone() here, do you?
|
@nikomatsakis Fixed nits.
Do you want me to work on that in this PR? |
|
You can leave it as it is currently, however, if you can add a FIXME of some sort to that effect. |
|
Ok. Should I tidy the branch up too while I'm in there? |
|
Sure! |
wesleywiser
force-pushed
the
mir_switch
branch
from
65231f8 to
2b15361
Compare
|
I've squashed all my changes and rebased against master. |
|
@bors r+ nice! |
|
📌 Commit 2b15361 has been approved by |
Fixes #29574