[ty] Infer parameter specializations of generic aliases #18021
Merged
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
|
AlexWaygood
reviewed
May 12, 2025
Comment on lines
50
to
52
| // Keep this field private, so that the only way of constructing `NominalInstanceType` instances | ||
| // is through the `Type::instance` constructor function. | ||
| class: ClassType<'db>, | ||
| pub(super) class: ClassType<'db>, |
Member
There was a problem hiding this comment.
hmm, the comment here says to keep this field private ;)
Member
Author
There was a problem hiding this comment.
Woof, talk about tunnel vision! I had done this to be able to pattern-match on the field so that I could have a match arm that only applied to nominal instances of generic aliases.
I'll back this out and tweak the match arm — though keeping it that way might become more difficult in the future if/when the match becomes more complex, with fallthroughs being needed.
Member
Author
There was a problem hiding this comment.
Oh, actually there's another trick we can use here. Let me push it up as a proposal to see what you think
Member
Author
There was a problem hiding this comment.
Here you go — if we add an unused PhantomData field, we can mark that as private to prevent construction, but still allow read/match access to the class field. wdyt? (If you're 👍 then I'll also remove the class() accessor method in favor of just reading the field, but didn't want to clutter the diff with that yet if you prefer the original)
Member
There was a problem hiding this comment.
Yeah, I like the way that it allows for easier pattern-matching while preserving the property that it's only possible to construct NominalInstanceTypes from the Type::instance() method!
AlexWaygood
reviewed
May 12, 2025
Comment on lines
676
to
683
| ( | ||
| Type::NominalInstance(NominalInstanceType { | ||
| class: ClassType::Generic(formal_alias), | ||
| }), | ||
| Type::NominalInstance(NominalInstanceType { | ||
| class: ClassType::Generic(actual_alias), | ||
| }), | ||
| ) if formal_alias.origin(self.db) == actual_alias.origin(self.db) => { |
Member
There was a problem hiding this comment.
what about generic protocol instances? Do we need to add a branch for them as well?
Member
Author
There was a problem hiding this comment.
Good idea, though I'd like to tackle that as a follow-on PR
AlexWaygood
reviewed
May 12, 2025
Comment on lines
57
to
+59
| // Keep this field private, so that the only way of constructing `NominalInstanceType` instances | ||
| // is through the `Type::instance` constructor function. | ||
| class: ClassType<'db>, | ||
| _phantom: PhantomData<()>, |
Member
There was a problem hiding this comment.
oh, that's clever! It's a shame I'm only learning about this trick now -- much of the diff in #17525 came from me making the class field private, and I guess that was unnecessary
* main: disable jemalloc on android (#18033) [ty] Fix incorrect type of `src.root` in documentation (#18040) [ty] Refine message for why a rule is enabled (#18038) [ty] Remove brackets around option names (#18037) Update pre-commit dependencies (#18025) Update docker/build-push-action action to v6.16.0 (#18030) Update docker/login-action action to v3.4.0 (#18031) Update taiki-e/install-action digest to 83254c5 (#18022) Update cargo-bins/cargo-binstall action to v1.12.4 (#18023) Update Rust crate ctrlc to v3.4.7 (#18027) Update Rust crate clap to v4.5.38 (#18026) Update Rust crate jiff to v0.2.13 (#18029) Update Rust crate getrandom to v0.3.3 (#18028) Update dependency ruff to v0.11.9 (#18024) [`pylint`] add fix safety section (`PLW1514`) (#17932) python_stdlib: update for 3.14 (#18014) [`ruff`] add fix safety section (`RUF033`) (#17760) [`pylint`] add fix safety section (`PLC0414`) (#17802)
This was referenced May 12, 2025
AlexWaygood
approved these changes
May 12, 2025
carljm
approved these changes
May 12, 2025
Contributor
carljm
left a comment
carljm
left a comment
There was a problem hiding this comment.
We adds the parentheses, then we takes the parentheses away again.
(Also, the actual fix here is excellent)
…eepish * origin/main: [ty] Induct into instances and subclasses when finding and applying generics (#18052) [ty] Allow classes to inherit from `type[Any]` or `type[Unknown]` (#18060) [ty] Allow a class to inherit from an intersection if the intersection contains a dynamic type and the intersection is not disjoint from `type` (#18055) [ty] Narrowing for `hasattr()` (#18053) Update reference documentation for `--python-version` (#18056) [`flake8-bugbear`] Ignore `B028` if `skip_file_prefixes` is present (#18047) [`airflow`] Apply try-catch guard to all AIR3 rules (`AIR3`) (#17887) [`pylint`] add fix safety section (`PLW3301`) (#17878) Update `--python` to accept paths to executables in virtual environments (#17954) [`pylint`] add fix safety section (`PLE4703`) (#17824) [`ruff`] Implement a recursive check for `RUF060` (#17976) [`flake8-use-pathlib`] `PTH*` suppress diagnostic for all `os.*` functions that have the `dir_fd` parameter (#17968) [`refurb`] Mark autofix as safe only for number literals in `FURB116` (#17692) [`flake8-simplify`] Fix `SIM905` autofix for `rsplit` creating a reversed list literal (#18045) Avoid initializing progress bars early (#18049)
…eepish * origin/main: [ty] Understand homogeneous tuple annotations (#17998)
dcreager
added a commit
that referenced
this pull request
May 13, 2025
…eep-dish * origin/main: [ty] Infer parameter specializations of generic aliases (#18021) [ty] Understand homogeneous tuple annotations (#17998) [ty] Induct into instances and subclasses when finding and applying generics (#18052) [ty] Allow classes to inherit from `type[Any]` or `type[Unknown]` (#18060) [ty] Allow a class to inherit from an intersection if the intersection contains a dynamic type and the intersection is not disjoint from `type` (#18055) [ty] Narrowing for `hasattr()` (#18053) Update reference documentation for `--python-version` (#18056) [`flake8-bugbear`] Ignore `B028` if `skip_file_prefixes` is present (#18047) [`airflow`] Apply try-catch guard to all AIR3 rules (`AIR3`) (#17887) [`pylint`] add fix safety section (`PLW3301`) (#17878) Update `--python` to accept paths to executables in virtual environments (#17954) [`pylint`] add fix safety section (`PLE4703`) (#17824) [`ruff`] Implement a recursive check for `RUF060` (#17976) [`flake8-use-pathlib`] `PTH*` suppress diagnostic for all `os.*` functions that have the `dir_fd` parameter (#17968) [`refurb`] Mark autofix as safe only for number literals in `FURB116` (#17692) [`flake8-simplify`] Fix `SIM905` autofix for `rsplit` creating a reversed list literal (#18045) Avoid initializing progress bars early (#18049)
dcreager
added a commit
that referenced
this pull request
May 13, 2025
…c protocols (#18054) Follows on from (and depends on) #18021. This updates our function specialization inference to infer type mappings from parameters that are generic protocols. For now, this only works when the argument _explicitly_ implements the protocol by listing it as a base class. (We end up using exactly the same logic as for generic classes in #18021.) For this to work with classes that _implicitly_ implement the protocol, we will have to check the types of the protocol members (which we are not currently doing), so that we can infer the specialization of the protocol that the class implements. --------- Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
Glyphack
pushed a commit
to Glyphack/ruff
that referenced
this pull request
May 21, 2025
) This updates our function specialization inference to infer type mappings from parameters that are generic aliases, e.g.: ```py def f[T](x: list[T]) -> T: ... reveal_type(f(["a", "b"])) # revealed: str ``` Though note that we're still inferring the type of list literals as `list[Unknown]`, so for now we actually need something like the following in our tests: ```py def _(x: list[str]): reveal_type(f(x)) # revealed: str ```
Glyphack
pushed a commit
to Glyphack/ruff
that referenced
this pull request
May 21, 2025
…c protocols (astral-sh#18054) Follows on from (and depends on) astral-sh#18021. This updates our function specialization inference to infer type mappings from parameters that are generic protocols. For now, this only works when the argument _explicitly_ implements the protocol by listing it as a base class. (We end up using exactly the same logic as for generic classes in astral-sh#18021.) For this to work with classes that _implicitly_ implement the protocol, we will have to check the types of the protocol members (which we are not currently doing), so that we can infer the specialization of the protocol that the class implements. --------- Co-authored-by: Alex Waygood <Alex.Waygood@Gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
This updates our function specialization inference to infer type mappings from parameters that are generic aliases, e.g.:
Though note that we're still inferring the type of list literals as
list[Unknown], so for now we actually need something like the following in our tests: