[ty] Avoid redundant constraint saturation work#25786
Merged
Merged
Conversation
Typing conformance resultsNo changes detected ✅Current numbersThe percentage of diagnostics emitted that were expected errors held steady at 94.36%. The percentage of expected errors that received a diagnostic held steady at 88.82%. The number of fully passing files held steady at 93/134. |
Memory usage reportSummary
Significant changesClick to expand detailed breakdownflake8
trio
sphinx
prefect
|
|
28d7fdb to
be0c908
Compare
c433edc to
294e3f1
Compare
dcreager
reviewed
Jun 9, 2026
be0c908 to
ab9db7e
Compare
294e3f1 to
d249349
Compare
ab9db7e to
6fd42b6
Compare
Base automatically changed from
charlie/fix-dynamic-constraint-explosion
to
main
June 10, 2026 20:23
d249349 to
a9a2a50
Compare
charliermarsh
added a commit
that referenced
this pull request
Jun 10, 2026
## Summary PR #25786 originally cached repeated constraint-set subtype checks on `ConstraintSetBuilder`. [During review](#25786 (comment)), that cache was replaced with a Salsa-tracked query to allow reuse across builders. On pytest-robotframework, the tracked query does not recover the same performance: current main remains roughly 25% slower than the commit immediately before #25778. This restores the builder-local `FxHashMap` cache and routes the two transitive-pivot subtype checks through it. The separate type-variable-bound precheck from #25786 remains unchanged. In 80 balanced, interleaved runs of pytest-robotframework using clean profiling builds, the results were: | Revision | Median | Paired change vs. pre-#25778 | | --- | ---: | ---: | | Pre-#25778 (`1676ec8013`) | 103.2 ms | — | | Current main (`8f2caa6a04`) | 128.3 ms | +24.9% | | This PR | 106.9 ms | +3.7% | This PR is 16.9% faster than current main. All three revisions produced byte-for-byte identical diagnostics. The full `ty_python_semantic` suite (558 tests), focused Clippy, and repository hooks pass.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Summary
The constraint-set subtype checks introduced in #25778 can be repeated many times while saturating pairs of constraints. On pytest-robotframework, that increased median check time from 105.8 ms on main to 159.9 ms with #25778.
This caches those checks for the lifetime of a constraint-set builder.
As a separate, semantics-preserving optimization, this also skips calling
add_nested_typevar_sequentswhen neither constraint has a type variable in its bounds. In that case, the method cannot derive any sequents: all of its variance checks returnBivariant. The precheck visits lazy types to match the traversal performed byvariance_of.With this change, pytest-robotframework takes 108.5 ms, 2.6% above main, with byte-for-byte identical diagnostics. The full
ty_python_semantictest suite, focused Clippy, and repository hooks pass.