rustdoc: improve crate-local inlining #32142
Conversation
mitaa
force-pushed
the
rdoc-maybe-inline-local
branch
2 times, most recently
from
91ea7c7 to
1953f4b
Compare
| while let Some(id) = cx.map.get_enclosing_scope(node) { | ||
| node = id; | ||
| if !seen.insert(id) { | ||
| break; |
There was a problem hiding this comment.
I'm not quite sure why this is necessary. (rustdoc/viewpath-self.rs is caught in an infinite loop otherwise)
mitaa
force-pushed
the
rdoc-maybe-inline-local
branch
2 times, most recently
from
7792679 to
18506c8
Compare
|
Thanks @mitaa! I wonder if we might be able to get away from dealing with the passes for now? Is there a use case where rustdoc is actually being called with various passes and this isn't working? I tend to view the passes in rustdoc as kinda a half-baked feature which wasn't ever really ready for prime time :( |
mitaa
force-pushed
the
rdoc-maybe-inline-local
branch
from
18506c8 to
7c98399
Compare
|
I've backed that commit out for now. The only thing rustdoc is doing "wrong" is overzealous inlining when it isn't required. It is definitely unfortunate that some passes have to currently to leak out like that such that various points in rustdoc need to have knowledge about them hardcoded in. |
This is actually a reexported implementation detail in the `rt::v1` module but rustdoc doesn't like reexporting items from `doc(hidden)` modules. Do what we'd end up doing anyway from an API perspective and make a public-facing `enum` that can be retranslated under the hood if necessary.
…excrichton rustdoc: improve crate-local inlining fixes rust-lang#28537 r? @alexcrichton
rustdoc: improve crate-local inlining fixes #28537 r? @alexcrichton
fixes #28537
r? @alexcrichton