[mir-opt] simplify Repeats that don't actually repeat the operand
#135322
Conversation
|
r? @Nadrieril rustbot has assigned @Nadrieril. Use |
|
Some changes occurred to MIR optimizations cc @rust-lang/wg-mir-opt |
This comment has been minimized.
This comment has been minimized.
b1a6483 to
699703e
Compare
| // CHECK-LABEL: fn remove_generic_array | ||
| // CHECK: debug a => const ZeroSized: [T; 0]; | ||
| // CHECK-NOT: = []; | ||
| let a = [x; 0]; |
There was a problem hiding this comment.
Also add a test where the zero comes from a constant not a literal
| // CHECK: _0 = [move {{_[0-9]+}}]; | ||
| // CHECK-NOT: [move {{_[0-9]+}}; 1] | ||
|
|
||
| [x; 1] |
There was a problem hiding this comment.
Also test getting the 1 from a constant, not a literal. I think the normalization for that has already happened, but we've had some issues with that recently
| // unreachable or can't be ZST | ||
| _ => false, | ||
| _ => Some(false), |
There was a problem hiding this comment.
exhaustively match here. This branch must not have false positives given that it means "definitely not"
There was a problem hiding this comment.
Good point. Spurious false is benign in how it's actually used, but that's not what the comments say, so I should re-work it. Maybe just have the wildcard be _ => None, since that's more obviously fine, and add the known-false cases explicitly.
|
☔ The latest upstream changes (presumably #135274) made this pull request unmergeable. Please resolve the merge conflicts. |
Created because when I was writing this case in GVN
https://github.com/rust-lang/rust/pull/133324/files#diff-292b215fdc6b59e3f3c4173c2270b14591c5673832fbfb05cd69a05c2ef0c30eR977-R979
I happened to notice that it worked for
[x]but not for[x; 1], so figured it would be good to simplify thatRepeatto the simplerAggregate.