remove support for `typeof` #148256

lcnr · 2025-10-29T15:57:19Z

see rust-lang/compiler-team#940 closes #148700

This also enables checks for invariants previously broken by typeof again.

r? types

WaffleLapkin · 2025-10-29T16:37:02Z

compiler/rustc_ast/src/token.rs

    !ident_token.is_reserved_ident()
        || ident_token.is_path_segment_keyword()
-        || [kw::Underscore, kw::For, kw::Impl, kw::Fn, kw::Unsafe, kw::Extern, kw::Typeof, kw::Dyn]
+        || [kw::Underscore, kw::For, kw::Impl, kw::Fn, kw::Unsafe, kw::Extern, kw::Dyn]


q: is there a chance this affects macros?

Yes, the following artificial snippet will go from pass→fail:

macro_rules! ck { ($ty:ty) => {}; } ck!(typeof(0));

After this PR it'll fail with no rules expected keyword `typeof` (and hypothetically if typeof were to be demoted to a normal identifier, it would start failing with expected type, found `0`).

However, it's very unlikely anybody is relying on it. So it's not really a concern. MBE is notoriously forward incompatible.

WaffleLapkin · 2025-10-29T16:40:04Z

compiler/rustc_hir_analysis/src/coherence/inherent_impls.rs

+            | ty::CoroutineWitness(..)
+            | ty::Alias(ty::Free, _)
+            | ty::Bound(..)
+            | ty::Placeholder(_)
+            | ty::Infer(_) => {


q: How is this related to this PR?

With the removal of typeof inference, it should now be impossible to sneak these unnameable types into an impl (even with TAIT you can't do that IINM). The typeof removal is partially motivated by being able to perform this exact change as lcnr alluded to in the MCP. CC #147219, #146620 #147146.

With the removal of typeof inference, it should now be impossible to sneak these unnameable types into an impl (even with TAIT you can't do that IINM)

Yes, there's currently no way to get unnameable types into impl headers or encounter them in coherence, and it's nice to be able to reason about that

Noratrieb · 2025-10-29T16:43:14Z

compiler/rustc_parse/src/parser/ty.rs

            // Reference
            self.expect_and()?;
            self.parse_borrowed_pointee()?
-        } else if self.eat_keyword_noexpect(kw::Typeof) {


It may be nice to keep a precise error message here and recover into an error type

how do you do that? It's not immediately obvious to me how to impl that and don't want to spend much time on figuring this out 🤔

how do you do that? It's not immediately obvious to me how to impl that

Instead of removing this eat_keyword_noexpect branch + fn parse_typeof_ty, keep both but change parse_typeof_ty to no longer return a TyKind::Typeof(expr) but instead a TyKind::Err(guar) (this variant already exists) where obv. let guar = dcx.struct_span_err(…, "…");.

This made me realize that we do accept this on stable (and have been since ~forever), lol:

#[cfg(false)] type X = typeof(0);

So forget what I said about MBEs, this is a far more obvious demonstration why removing typeof is a theoretical breaking change, not that anybody would care ^^.

done that I think? @fmease do you think this is relevant enough to do a crater run or sth like that here?

I think the most likely scenario would be something like #[mydsl] fn f() { … typeof(…) … } but even that seems farfetched (esp. since this would still compile if mydsl is a proc-macro attr).

I've also skimmed the results of GH Search: /(?-i)\btypeof\([^)"]/ language:Rust but couldn't find anything that would break.

So, I don't think this warrants a crater run, let's just ship it.

spastorino · 2025-11-06T19:59:55Z

This looks good to me, r=me once the MCP finishes and addressing @Noratrieb's comment.

rustbot · 2025-11-25T09:19:44Z

HIR ty lowering was modified

cc @fmease

rustbot · 2025-11-25T09:19:58Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

rustbot · 2025-11-25T10:34:35Z

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

Some changes occurred in src/tools/rustfmt

cc @rust-lang/rustfmt

rust-log-analyzer · 2025-11-25T17:22:11Z

The job aarch64-gnu-llvm-20-1 failed! Check out the build log: (web) (plain enhanced) (plain)

Click to see the possible cause of the failure (guessed by this bot)


failures:

---- [ui] tests/rustdoc-ui/issues/ice-typeof-102986.rs stdout ----
Saved the actual stderr to `/checkout/obj/build/aarch64-unknown-linux-gnu/test/rustdoc-ui/issues/ice-typeof-102986/ice-typeof-102986.stderr`
diff of stderr:

2   --> $DIR/ice-typeof-102986.rs:3:9
3    |
4 LL |     y: (typeof("hey"),),
-    |         ^^^^^^^^^^^^^ reserved keyword
+    |         ^^^^^^^^^^^^^
6    |
- help: consider replacing `typeof(...)` with an actual type
-    |
- LL -     y: (typeof("hey"),),
- LL +     y: (&str,),
-    |
+    = note: consider replacing `typeof(...)` with an actual type
12 
- error: aborting due to 1 previous error
+ error: Compilation failed, aborting rustdoc
---
To only update this specific test, also pass `--test-args issues/ice-typeof-102986.rs`

error: 1 errors occurred comparing output.
status: exit status: 1
command: env -u RUSTC_LOG_COLOR RUSTC_ICE="0" RUST_BACKTRACE="short" "/checkout/obj/build/aarch64-unknown-linux-gnu/stage2/bin/rustdoc" "/checkout/tests/rustdoc-ui/issues/ice-typeof-102986.rs" "-Zthreads=1" "-Zsimulate-remapped-rust-src-base=/rustc/FAKE_PREFIX" "-Ztranslate-remapped-path-to-local-path=no" "-Z" "ignore-directory-in-diagnostics-source-blocks=/cargo" "-Z" "ignore-directory-in-diagnostics-source-blocks=/checkout/vendor" "--sysroot" "/checkout/obj/build/aarch64-unknown-linux-gnu/stage2" "--target=aarch64-unknown-linux-gnu" "--check-cfg" "cfg(test,FALSE)" "--error-format" "json" "--json" "future-incompat" "-Ccodegen-units=1" "-Zui-testing" "-Zdeduplicate-diagnostics=no" "-Zwrite-long-types-to-disk=no" "-Cstrip=debuginfo" "-o" "/checkout/obj/build/aarch64-unknown-linux-gnu/test/rustdoc-ui/issues/ice-typeof-102986" "-A" "internal_features" "-A" "unused_parens" "-A" "unused_braces" "-Cdebuginfo=0"
stdout: none
--- stderr -------------------------------
error[E0516]: `typeof` is a reserved keyword but unimplemented
##[error]  --> /checkout/tests/rustdoc-ui/issues/ice-typeof-102986.rs:3:9
   |
LL |     y: (typeof("hey"),),
   |         ^^^^^^^^^^^^^
   |
   = note: consider replacing `typeof(...)` with an actual type

error: Compilation failed, aborting rustdoc

rustbot assigned spastorino Oct 29, 2025

lcnr marked this pull request as ready for review October 29, 2025 15:57

lcnr force-pushed the orphan-check branch from e0214f1 to d1fb1a3 Compare October 29, 2025 16:17

lcnr changed the title ~~orphan check: don't support unnameable types~~ remove support for typeof Oct 29, 2025

lcnr mentioned this pull request Oct 29, 2025

remove support for typeof rust-lang/compiler-team#940

Closed

3 tasks

lcnr added the S-waiting-on-MCP Status: PR has a compiler MCP and is waiting for the compiler MCP to complete. label Oct 29, 2025

traviscross added the I-lang-radar Items that are on lang's radar and will need eventual work or consideration. label Oct 29, 2025

WaffleLapkin reviewed Oct 29, 2025

View reviewed changes

Noratrieb reviewed Oct 29, 2025

View reviewed changes

cyrgani linked an issue Oct 29, 2025 that may be closed by this pull request

ICE on typeof(CT_PARAM) due to incorrect generic param substitution / instantiation somewhere #146384

Open

lcnr force-pushed the orphan-check branch from d1fb1a3 to 5c3a019 Compare November 25, 2025 09:19

remove support for type-of

feb1303