Revert PR 81473 to resolve (on mainline) issues 81626 and 81658.#86212
Conversation
|
(rust-highfive has picked a reviewer for you, use r? to override) |
This comment has been minimized.
This comment has been minimized.
|
r=me but looks like there's CI trouble here, and presumably we'll need another backport as this unfortunately didn't land in time for 1.54 branching |
|
Triage: |
|
Beta-nominated for 1.54, as this will land in 1.55. |
|
☔ The latest upstream changes (presumably #87129) made this pull request unmergeable. Please resolve the merge conflicts. |
|
Hmm. Do I delete the unused method, or just |
I expect these two methods to come back very soon; noise of removing them to satisfy lint seems wrong.
7ae061c to
b6e9d06
Compare
|
@bors r=simulacrum also, unilaterally approving for beta backport. |
|
📌 Commit ee2bb1f has been approved by |
|
@bors p=1 |
|
☀️ Test successful - checks-actions |
…warn-write-only-fields, r=Mark-Simulacrum Beta targetted revert 81473 warn write only fields beta backport of PR rust-lang#86212
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
…, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (rust-lang#81626, rust-lang#81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in rust-lang#87129 to address rust-lang#75356 - The implementation was disabled alongside the revert of rust-lang#81473's "write-only fields" detection - rust-lang#81473 was reverted via rust-lang#86212 and rust-lang#83171 due to false positives in write-only field detection (rust-lang#81626, rust-lang#81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
Rollup merge of #145214 - notJoon:fix/enable-self-assignment, r=petrochenkov fix: re-enable self-assignment ## Description Re-enables the self-assignment detection that was previously disabled due to unrelated regressions. The fix detects useless assignments like `x = x` and `foo.field = foo.field`. ## History The original regressions (#81626, #81658) were specifically about false positives in write-only field detection, not self-assignment detection. Belows are brief history for the rule that I understand. - Self-assignment detection was originally implemented in #87129 to address #75356 - The implementation was disabled alongside the revert of #81473's "write-only fields" detection - #81473 was reverted via #86212 and #83171 due to false positives in write-only field detection (#81626, #81658) - The self-assignment detection feature got removed, even though it wasn't the reason for the problems This PR only re-enables the self-assignment checks, which are orthogonal to the problematic write-only field analysis. ## Changes - Removed `#[allow(dead_code)]` from `compiler/rustc_passes/src/dead.rs` file - `handle_assign` and - `check_for_self_assign` - Added `ExprKind::Assign` handling in `visit_expr` to call both methods - Updated test expectations in `tests/ui/lint/dead-code/self-assign.rs`
This is a nightly-targetted variant of PR #83171
The intent is to just address issue #81658 on all release channels, rather that keep repeatedly reverting PR #83171 on beta.
However, our intent is also to reland PR #83171 after we have addressed issue #81658 , most likely by coupling the re-landing of PR #83171 with an enhancement like PR #83004