[red-knot] Create generic context for generic classes lazily #17617
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
|
MichaReiser
approved these changes
Apr 25, 2025
Member
MichaReiser
left a comment
MichaReiser
left a comment
There was a problem hiding this comment.
This looks nice!
It does lead to a small 3% perf regression on incremental benchmarks. I suspect it is because it now requires two salsa ingredients for every class (generic context and the class itself).
* main: (24 commits) [`airflow`] Extend `AIR301` rule (#17598) [`airflow`] update existing `AIR302` rules with better suggestions (#17542) red_knot_project: sort diagnostics from checking files [red-knot] fix detecting a metaclass on a not-explicitly-specialized generic base (#17621) [red-knot] fix inheritance-cycle detection for generic classes (#17620) [`pylint`] Detect `global` declarations in module scope (`PLE0118`) (#17411) Add Semantic Error Test for LateFutureImport (#17612) [red-knot] change TypeVarInstance to be interned, not tracked (#17616) [red-knot] Special case `@final`, `@override` (#17608) [red-knot] add TODO comment in specialization code (#17615) [`semantic-syntax-errors`] test for `LoadBeforeGlobalDeclaration` - ruff linter (#17592) [syntax-errors] `nonlocal` declaration at module level (#17559) [`airflow`] Apply auto fix to cases where name has been changed in Airflow 3 (`AIR311`) (#17571) [syntax-errors] Make `async-comprehension-in-sync-comprehension` more specific (#17460) Bump 0.11.7 (#17613) [red-knot] Use iterative approach to collect overloads (#17607) red_knot_python_semantic: avoid Rust's screaming snake case convention in mdtest red_knot_python_semantic: improve diagnostics for unsupported boolean conversions red_knot_python_semantic: add "return type span" helper method red_knot_python_semantic: move parameter span helper method ...
dcreager
added a commit
that referenced
this pull request
Apr 26, 2025
* main: (27 commits) [red-knot] Add new property tests for subtyping with "bottom" callable (#17635) [red-knot] Create generic context for generic classes lazily (#17617) ruff_db: add tests for annotations with no ranges [`airflow`] Extend `AIR301` rule (#17598) [`airflow`] update existing `AIR302` rules with better suggestions (#17542) red_knot_project: sort diagnostics from checking files [red-knot] fix detecting a metaclass on a not-explicitly-specialized generic base (#17621) [red-knot] fix inheritance-cycle detection for generic classes (#17620) [`pylint`] Detect `global` declarations in module scope (`PLE0118`) (#17411) Add Semantic Error Test for LateFutureImport (#17612) [red-knot] change TypeVarInstance to be interned, not tracked (#17616) [red-knot] Special case `@final`, `@override` (#17608) [red-knot] add TODO comment in specialization code (#17615) [`semantic-syntax-errors`] test for `LoadBeforeGlobalDeclaration` - ruff linter (#17592) [syntax-errors] `nonlocal` declaration at module level (#17559) [`airflow`] Apply auto fix to cases where name has been changed in Airflow 3 (`AIR311`) (#17571) [syntax-errors] Make `async-comprehension-in-sync-comprehension` more specific (#17460) Bump 0.11.7 (#17613) [red-knot] Use iterative approach to collect overloads (#17607) red_knot_python_semantic: avoid Rust's screaming snake case convention in mdtest ...
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As discussed today, this is needed to handle legacy generic classes without having to infer the types of the class's explicit bases eagerly at class construction time. Pulling this out into a separate PR so there's a smaller diff to review.
This also makes our representation of generic classes and functions more consistent — before, we had separate Rust types and enum variants for generic/non-generic classes, but a single type for generic functions. Now we each a single (respective) type for each.
There were very few places we were differentiation between generic and non-generic class literals, and these are handled now by calling the (salsa cached)
generic_contextaccessor function.Note that
ClassTypeis still an enum with distinct variants for non-generic classes and specialized generic classes.