Add `min_adt_const_params` gate by zedddie · Pull Request #153592 · rust-lang/rust

zedddie · 2026-03-09T05:36:09Z

Add min_adt_const_params feature gate to disallow ConstParamTy_ impl for types which fields visibility mismatches with struct itself

r? BoxyUwU

rustbot · 2026-03-09T05:36:14Z

BoxyUwU is currently at their maximum review capacity.
They may take a while to respond.

zedddie · 2026-03-09T06:15:39Z

compiler/rustc_feature/src/unstable.rs

    /// Allows using the `unadjusted` ABI; perma-unstable.
    (internal, abi_unadjusted, "1.16.0", None),
+    /// Restrict adt_const_params fields to share same privacy.
+    (internal, adt_const_params_restricted_privacy, "CURRENT_RUSTC_VERSION", None),


I am really not sure on whether this should be internal or unstable (I guess should share issue number with adt_const_params, so it would be unstable then)

this can be unstable. the difference there is generally about whether we expect users to actually use this feature. internal is for when users arent supposed to use it, but unstable is for stuff we expect/hope/want to stabilize eventually.

BoxyUwU

thanks this looks good! can you change the feature to be min_adt_const_params and have it not depend on adt_const_params. i expect this will likely require updating any features().adt_const_params() feature checks across the compiler to also check for min_adt_const_params()

sorry for sending you down a different path initially

View changes since this review

zedddie · 2026-03-18T12:38:22Z

compiler/rustc_feature/src/unstable.rs

    (unstable, marker_trait_attr, "1.30.0", Some(29864)),
    /// Enable mgca `type const` syntax before expansion.
    (incomplete, mgca_type_const_syntax, "1.95.0", Some(132980)),
+    /// Restrict adt_const_params fields to share same privacy.


not sure about what should I write here as it will be a standalone feature

I would do similar wording to what's on the adt_const_params entry, so maybe:

Allows additional const parameter types, such as [u8; 10] or user defined types. User defined types must not have fields more private than the type itself.

BoxyUwU · 2026-03-18T13:45:40Z

compiler/rustc_feature/src/unstable.rs

    /// Enable mgca `type const` syntax before expansion.
    (incomplete, mgca_type_const_syntax, "1.95.0", Some(132980)),
+    /// Restrict adt_const_params fields to share same privacy.
+    (unstable, min_adt_const_params, "CURRENT_RUSTC_VERSION", Some(95174)),


Suggested change

(unstable, min_adt_const_params, "CURRENT_RUSTC_VERSION", Some(95174)),

(unstable, min_adt_const_params, "CURRENT_RUSTC_VERSION", Some(154042)),

I've made a new tracking issue for this since we'll probably want a separate list of unresolved questions and whatnot 🤔

BoxyUwU · 2026-03-18T13:49:28Z

compiler/rustc_hir_analysis/src/coherence/builtin.rs

+                let struct_vis = tcx.visibility(adt.did());
+                for variant in adt.variants() {
+                    for field in &variant.fields {
+                        if field.vis != struct_vis {


Can you add the following as a test:

#![feature(min_adt_const_params)] use core::marker::ConstParamTy; #[derive(ConstParamTy, Eq, PartialEq)] struct Foo { pub field: u32, }

I think this will fail under your PR due to using equality instead of checking fields are "at least as visible as the struct". I'm not really sure what the desired behaviour is but it'll need to figured out as part of the RFC and having a test which demonstrates the choice is useful.

yes it will fail I think I tested this earlier, idk why I don't have this in test though. ~~Will add to existing one~~

I see its better as the distinct test as it will show existing test doesn't require adt_const_params

BoxyUwU · 2026-03-18T13:50:01Z

implementation looks good to me, thanks for working on this :)

BoxyUwU · 2026-03-18T13:52:01Z

compiler/rustc_hir_analysis/src/coherence/builtin.rs

        return Ok(());
    }

+    if tcx.features().min_adt_const_params() {


I guess this should actually be if tcx.features().min_adt_const_params() && !tcx.features().adt_const_params() so that once we stabilize min_adt_const_params you can avoid the restriction by enabling full adt const params

Can you add a test that the following compiles:

#![feature(min_adt_const_params, adt_const_params)] use core::marker::ConstParamTy; #[derive(ConstParamTy, Eq, PartialEq)] enum Foo { pub field: u32, }

BoxyUwU · 2026-03-23T17:47:26Z

compiler/rustc_hir_analysis/src/coherence/builtin.rs

        return Ok(());
    }

+    if tcx.features().min_adt_const_params() && !tcx.features().adt_const_params() {


lol yeah I guess I didn't think deeply enough about the fact that this means you can implement ConstParamTy for arbitrary stuff if you don't enable any feature gate x)

Suggested change

if tcx.features().min_adt_const_params() && !tcx.features().adt_const_params() {

if !tcx.features().adt_const_params() {

I'll r+ once this is done

Make sense, although even if they would implement ConstParamTy_ for their arbitrary struct, they couldn't use it anyway without min_adt_const_params or adt_const_params :> I am playing with tests a bit to include we ensurew this also without any gate

in theory you could have an unstable crate without min_adt_const_params and then a downstream crate which does enable it and can observe the broken impl and use it 🤔

but if they would implement it with adt_const_params, they still can use this unrestricted impl with min_adt_const_params on another crate 🧐

aux:

#![allow(incomplete_features)] #![feature(adt_const_params, const_param_ty_trait)] use std::marker::ConstParamTy_; #[derive(PartialEq, Eq)] pub struct Fumo { cirno: i32, pub(crate) reimu: i32 } impl ConstParamTy_ for Fumo {}

test:

//@aux-build:cross-crate-const_param_ty-impl-leak.rs #![allow(incomplete_features)] #![feature(min_adt_const_params)] extern crate cross_crate_const_param_ty_impl_leak; use cross_crate_const_param_ty_impl_leak::Fumo; fn fumo_fn<const N: Fumo>() {} fn main() {}

this passes for example

hmm yeah thats kinda weird lol. though it doesn't feel as bad to me because atleast someone enabled the feature gate for supporting private fields in const generics

eh I would add this test but it seems aux isn't supposed to fail -_-

Nadrieril · 2026-03-29T18:19:16Z

It happens 😁

BoxyUwU

this looks good to me. can you move the tests to be tests/ui/const-generics/min_adt_const_params rather than tests/ui/const-generics/adt_const_params/min_adt_const_params then I'll r+ :3

View changes since this review

BoxyUwU · 2026-04-03T10:27:21Z

@bors r+

thx

rust-bors · 2026-04-03T10:27:25Z

📌 Commit 72fbd1c has been approved by BoxyUwU

It is now in the queue for this repository.

…d_privacy-gate, r=BoxyUwU Add `min_adt_const_params` gate Add `min_adt_const_params` feature gate to disallow `ConstParamTy_` impl for types which fields visibility mismatches with struct itself r? BoxyUwU

Rollup of 5 pull requests Successful merges: - #153286 (various fixes for scalable vectors) - #153592 (Add `min_adt_const_params` gate) - #154051 (use libm for acosh and asinh for f16, f32, and f64) - #154743 (Remove an unused `StableHash` impl.) - #154752 (Add comment to borrow-checker)

Rollup of 7 pull requests Successful merges: - #153286 (various fixes for scalable vectors) - #153592 (Add `min_adt_const_params` gate) - #154675 (Improve shadowed private field diagnostics) - #154653 (Remove rustc_on_unimplemented's append_const_msg) - #154743 (Remove an unused `StableHash` impl.) - #154752 (Add comment to borrow-checker) - #154764 (Add tests for three ICEs that have already been fixed)

…d_privacy-gate, r=BoxyUwU Add `min_adt_const_params` gate Add `min_adt_const_params` feature gate to disallow `ConstParamTy_` impl for types which fields visibility mismatches with struct itself r? BoxyUwU

Rollup of 10 pull requests Successful merges: - #154376 (Remove more BuiltinLintDiag variants - part 4) - #127534 (feat(core): impl Step for NonZero<u*>) - #153286 (various fixes for scalable vectors) - #153592 (Add `min_adt_const_params` gate) - #154675 (Improve shadowed private field diagnostics) - #154703 (Fix trailing comma in lifetime suggestion for empty angle brackets) - #154653 (Remove rustc_on_unimplemented's append_const_msg) - #154743 (Remove an unused `StableHash` impl.) - #154752 (Add comment to borrow-checker) - #154764 (Add tests for three ICEs that have already been fixed)

Rollup of 7 pull requests Successful merges: - #153286 (various fixes for scalable vectors) - #153592 (Add `min_adt_const_params` gate) - #154675 (Improve shadowed private field diagnostics) - #154653 (Remove rustc_on_unimplemented's append_const_msg) - #154743 (Remove an unused `StableHash` impl.) - #154752 (Add comment to borrow-checker) - #154764 (Add tests for three ICEs that have already been fixed)

Rollup merge of #153592 - zedddie:adt_const_params_restricted_privacy-gate, r=BoxyUwU Add `min_adt_const_params` gate Add `min_adt_const_params` feature gate to disallow `ConstParamTy_` impl for types which fields visibility mismatches with struct itself r? BoxyUwU

rustbot assigned BoxyUwU Mar 9, 2026

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Mar 9, 2026

zedddie commented Mar 9, 2026

View reviewed changes

BoxyUwU reviewed Mar 16, 2026

View reviewed changes

zedddie force-pushed the adt_const_params_restricted_privacy-gate branch from 04c868a to 2871ec1 Compare March 18, 2026 12:33

zedddie commented Mar 18, 2026

View reviewed changes

BoxyUwU mentioned this pull request Mar 18, 2026

Tracking Issue for feature(min_adt_const_params) #154042

Open

7 tasks

BoxyUwU reviewed Mar 18, 2026

View reviewed changes

zedddie changed the title ~~Add adt_const_params_restricted_privacy gate~~ Add min_adt_const_params gate Mar 18, 2026

BoxyUwU reviewed Mar 18, 2026

View reviewed changes

zedddie force-pushed the adt_const_params_restricted_privacy-gate branch 2 times, most recently from fcb960a to 2c27a37 Compare March 18, 2026 16:12

zedddie marked this pull request as ready for review March 22, 2026 19:16

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Mar 22, 2026

BoxyUwU reviewed Mar 23, 2026

View reviewed changes

zedddie force-pushed the adt_const_params_restricted_privacy-gate branch from 2c27a37 to 4222812 Compare March 25, 2026 20:23

This comment was marked as outdated.

Sign in to view

BoxyUwU reviewed Apr 2, 2026

View reviewed changes

zedddie added 2 commits April 2, 2026 17:10

add min_adt_const_params gate tests

98d259b

implement min_adt_const_params feature

72fbd1c

zedddie force-pushed the adt_const_params_restricted_privacy-gate branch from 11d5d14 to 72fbd1c Compare April 2, 2026 15:41

rust-bors bot added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Apr 3, 2026

matthiaskrgr mentioned this pull request Apr 3, 2026

Rollup of 5 pull requests #154757

Closed

matthiaskrgr mentioned this pull request Apr 3, 2026

Rollup of 7 pull requests #154773

Merged

jhpratt mentioned this pull request Apr 4, 2026

Rollup of 10 pull requests #154793

Closed

rust-bors bot merged commit a4d278d into rust-lang:main Apr 4, 2026
11 checks passed

rustbot added this to the 1.96.0 milestone Apr 4, 2026

	(unstable, min_adt_const_params, "CURRENT_RUSTC_VERSION", Some(95174)),
	(unstable, min_adt_const_params, "CURRENT_RUSTC_VERSION", Some(154042)),

	if tcx.features().min_adt_const_params() && !tcx.features().adt_const_params() {
	if !tcx.features().adt_const_params() {

Uh oh!

Conversation

zedddie commented Mar 9, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Mar 9, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

BoxyUwU left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zedddie Mar 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

BoxyUwU commented Mar 18, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

zedddie Mar 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment was marked as outdated.

Nadrieril commented Mar 29, 2026

Uh oh!

BoxyUwU left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

BoxyUwU commented Apr 3, 2026

Uh oh!

rust-bors bot commented Apr 3, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

zedddie commented Mar 9, 2026 •

edited by rustbot

Loading

BoxyUwU left a comment •

edited

Loading

zedddie Mar 18, 2026 •

edited

Loading

zedddie Mar 25, 2026 •

edited

Loading

BoxyUwU left a comment •

edited by rustbot

Loading