Find the largest niche when computing layouts #50860
Conversation
|
(rust_highfive has picked a reviewer for you, use r? to override) |
rust-highfive
added
the
S-waiting-on-review
src/librustc/ty/layout.rs
Outdated
| @@ -1674,40 +1685,56 @@ impl<'a, 'tcx, C> TyLayoutMethods<'tcx, C> for Ty<'tcx> | |||
| } | |||
| } | |||
|
|
|||
| pub struct Niche { | |||
There was a problem hiding this comment.
Can this struct remain private, for now?
Otherwise we end up with `Option<Option<(&(), bool)>>` unnecessarily large.
nox
force-pushed
the
big-niches-for-big-doggos-
branch
from
fbba4d9 to
b7db68f
Compare
|
@bors try cc @Mark-Simulacrum @nikomatsakis We should try to figure out how expensive this might be (it could result in type traversals doing much more work, and adding caching is non-trivial in this case). |
|
⌛ Trying commit b7db68f with merge 3efc4382f7de39b68b475b09666347b27368ee3f... |
|
☀️ Test successful - status-travis |
kennytm
added
the
S-waiting-on-perf
|
r? @eddyb |
|
Perf queued. |
|
Perf result is now available http://perf.rust-lang.org/compare.html?start=fd18d2537ddcffb24b3739393b085ed28dfc340e&end=3efc4382f7de39b68b475b09666347b27368ee3f&stat=instructions%3Au (stylo-servo's "clean incremental" result is spurious because of a hash table; all other results are within ±1.5%) |
kennytm
removed
the
S-waiting-on-perf
|
So this can land as is, right? |
|
Nice branch name 😄 🐶 |
|
@bors r+ |
|
📌 Commit b7db68f has been approved by |
bors
added
S-waiting-on-bors
Find the largest niche when computing layouts Otherwise we end up with `Option<Option<(&(), bool)>>` unnecessarily large.
|
☀️ Test successful - status-appveyor, status-travis |
Otherwise we end up with
Option<Option<(&(), bool)>>unnecessarily large.