chore(uv): update ty requirement from >=0.0.32 to >=0.0.33

[dependabot]

Updates the requirements on ty to permit the latest version.

Release notes

Sourced from ty's releases.

0.0.33

Release Notes

Released on 2026-04-28.

Notable changes

• ty now prefers the declared type of an annotated assignment in more situations (#24802). Consider this example:

  from some_library import untyped_function
  threshold: int | None = 0
  result: str = untyped_function()

  ty previously favored the inferred type of the right hand side expression when threshold and result were used. This is useful for threshold, as it allows something like threshold += 1 to work without an error: we know that threshold could later become None, but right now, we see that it is an int. However, for result, the inferred type is Unknown. This is not a useful type and it can lead to false negatives. Starting with this release, ty will therefore prefer the declared type if the inferred and declared types are mutually assignable. In the above example, threshold will still be inferred as int (or rather Literal[1]), but result will now be inferred as str. If you previously added casts to work around this behavior, you should be able to remove them after upgrading.

Bug fixes

• Fix reporting of annotation-only locals as unused (#24811)
• Fix project and workspace selection (#24824)
• Fix go-to definition for generic classes (#24714)
• Fix receiver coloring for aliased decorators (#24884)

LSP server

• Add support for go-to definition in literal enum member inlay hints (#24792)
• Add support for "baking" keyword argument inlay hints into the source code (#24667)
• Don't allow inlay hint edits when introducing a non global scope symbol (#24797)
• Omit semantic highlighting for unresolved symbols (#24718)

Core type checking

• Support narrowing with aliased conditional expressions (#24302)
• Model short-circuiting control flow in Boolean expressions (#24458)
• Handle finally blocks where all try/except blocks are terminal (#24882)
• Detect invalid ClassVar vs instance-attribute overrides (#24767)
• Emit diagnostic for invalid uses of Unpack[...] (#24868)
• Infer lambda parameter types with Callable type context (#24317)
• Support ** unpacking of TypedDict in dict-literal assignments (#24703)
• Support Unpack[TypedDict] in **kwargs signatures (#24653)
• Treat [*xs] as an irrefutable pattern when matching on Sequence (#24787)
• Improve generics solving for unions in invariant positions (#24698)
• Improve generics solving for unions when matching against protocols (#24837)

Diagnostics

... (truncated)

Changelog

Sourced from ty's changelog.

0.0.33

Released on 2026-04-28.

Notable changes

• ty now prefers the declared type of an annotated assignment in more situations (#24802). Consider this example:

  from some_library import untyped_function
  threshold: int | None = 0
  result: str = untyped_function()

  ty previously favored the inferred type of the right hand side expression when threshold and result were used. This is useful for threshold, as it allows something like threshold += 1 to work without an error: we know that threshold could later become None, but right now, we see that it is an int. However, for result, the inferred type is Unknown. This is not a useful type and it can lead to false negatives. Starting with this release, ty will therefore prefer the declared type if the inferred and declared types are mutually assignable. In the above example, threshold will still be inferred as int (or rather Literal[1]), but result will now be inferred as str. If you previously added casts to work around this behavior, you should be able to remove them after upgrading.

Bug fixes

• Fix reporting of annotation-only locals as unused (#24811)
• Fix project and workspace selection (#24824)
• Fix go-to definition for generic classes (#24714)
• Fix receiver coloring for aliased decorators (#24884)

LSP server

• Add support for go-to definition in literal enum member inlay hints (#24792)
• Add support for "baking" keyword argument inlay hints into the source code (#24667)
• Don't allow inlay hint edits when introducing a non global scope symbol (#24797)
• Omit semantic highlighting for unresolved symbols (#24718)

Core type checking

• Support narrowing with aliased conditional expressions (#24302)
• Model short-circuiting control flow in Boolean expressions (#24458)
• Handle finally blocks where all try/except blocks are terminal (#24882)
• Detect invalid ClassVar vs instance-attribute overrides (#24767)
• Emit diagnostic for invalid uses of Unpack[...] (#24868)
• Infer lambda parameter types with Callable type context (#24317)
• Support ** unpacking of TypedDict in dict-literal assignments (#24703)
• Support Unpack[TypedDict] in **kwargs signatures (#24653)
• Treat [*xs] as an irrefutable pattern when matching on Sequence (#24787)
• Improve generics solving for unions in invariant positions (#24698)
• Improve generics solving for unions when matching against protocols (#24837)

Diagnostics

• Add error context to invalid-return-type diagnostics, invalid-yield diagnostics, attribute assignment diagnostics (#24770, #24771)

... (truncated)

Commits

• c512d84 Bump version to 0.0.33 (#3368)
• 4cd7b33 Upgrade Depot runners from macOS 14 to 15 (#3363)
• c78b832 Update rui314/setup-mold digest to 9c9c13b (#3342)
• dea3381 Update actions/cache action to v5.0.5 (#3343)
• d451af4 update typing-features and faqs (#3335)
• 052d70b Update prek dependencies (#3344)
• 66b5e87 Update astral-sh/setup-uv action to v8.1.0 (#3345)
• 7ec6712 Add a 'Diagnostics improvements' section to the changelogs (#3309)
• 978dfdb Add version metadata publishing to the release process (#3292)
• See full diff in compare view

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (2c201e6) to head (2aeb23d).
⚠️ Report is 6 commits behind head on develop.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff            @@
##           develop      #232   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            9         9           
  Lines          103       103           
  Branches         1         1           
=========================================
  Hits           103       103           

Flag	Coverage Δ
unittests	100.00% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2c201e6...2aeb23d. Read the comment docs.