[ConstraintSystem] Implement lazy transitive supertype inference #86326
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
…with other type variables
There are two variants of `introduceToInference` now, one for a fixed type binding and the other is for a new constraint added to the node. This wasn't necessary before since it was a simple `infer` call on `PotentialBindings` but the plan is to expand this into performing the transitive inference through subtype chains as well.
This could be used by the constraint graph to propagate transtiive bindings through subtype chains.
…ariables This is important to perform actions and propagate bindings and and down subtype/supertype chains.
…type bindings A constraint graph node can now receive and propagate subtype bindings up its supertype chain. At the same time, given a set of constraints a constraint graph node can recursively, starting from itself, delete all the transitive bindings associated with them.
…erence` The `mergeNodesPre` effectively means the type variable that about to merged into an equivalence class should be removed from inference.
The functinality is enabled by `EnableTransitiveInference` flag. Instead of transferring supertype bindings up the chain during `determineBestBindings`, do it when new bindings become available. This avoids all of the ordering issues that plague the current approach and helps with the performance.
…ing transitive bindings This change enables the use of these methods for both subtype and supertype transitive bindings.
…upertypes If there are any loops in the graph current recursive approach cannot discover them because it only seems direct subtypes/supertypes but BFS algorithm that walks the whole chain starting for the given node can.
xedin
force-pushed
the
lazy-transitive-inference
branch
from
0cbd3d7 to
3bd6da1
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
The functionally is hidden behind a constraint system flag -
EnableTransitiveInferenceat the momentand can only be enabled for unit tests.
The constraint graph now tracks subtype/supertype chains established by introduction of
Subtype.When new exact/supertype binding for a type variable becomes available, it gets propagated up
the chain to all of the current supertypes of this type variable. Such "transitive" bindings get retracted
when a type variable gets bound or originator constraint gets retracted.
This is a replacement for the logic that currently lives in
determineBestBindingsand isn't currently in100% correct because it relies on the type variable ordering (i.e. it's wrong when a type variable with
a higher ID is a subtype of a type variable is a lower ID).