[ConstraintSystem] Add a type variable merging heuristic to addJoinConstraint #33296
Conversation
…nstraint. This heuristic merges type variables for literal expressions of the same kind. This is valid because such type variables will have the same set of constraints on them, and must be bound to the same type.
|
Still need to update the test and also use |
hborla
force-pushed
the
merge-joined-literal-typevars
branch
from
da40585 to
5e7d7a9
Compare
|
@swift-ci please smoke test compiler performance |
There was a problem hiding this comment.
I feel like finding a representative (and equivalence class) for expressions and forming a join are two separate operations. Maybe this could be split into addJoinConstraint which takes expressions and another overload of it which takes an iterator over type variables so one call call another when needed.
@xedin I don't think that, for this particular heuristic, joining types and merging representative type variables are two independent operations. Doing this in |
hborla
force-pushed
the
merge-joined-literal-typevars
branch
from
5e7d7a9 to
ad77613
Compare
|
@hborla Consider it from this perspective - join operator is responsive for establishing a particular relationship between a given set of type variables, its correctness is unrelated of properties of the type variables themselves. We are integrating heuristic into this method which is based on logic unrelated to the join operation itself - that all of the type variables of the same kind for an equivalence class - that by itself is true is certain situations e.g. if all of them are elements of a collection, arguments to a homogeneous call. I'm not sure if that's safe to do or does it just make it so any future use of |
|
@swift-ci please test source compatibility |
|
@swift-ci please smoke test compiler performance |
Summary for master smoketestNo regressions above thresholds Debugdebug briefRegressed (0)
Improved (0)
Unchanged (delta < 1.0% or delta < 100.0ms) (3)
debug detailedRegressed (0)
Improved (0)
Unchanged (delta < 1.0% or delta < 100.0ms) (16)
Releaserelease briefRegressed (0)
Improved (0)
Unchanged (delta < 1.0% or delta < 100.0ms) (3)
release detailedRegressed (0)
Improved (0)
Unchanged (delta < 1.0% or delta < 100.0ms) (16)
|
an iterator range and a callback to get the type. This allows callers to lazily compute the input types to join rather than constructing an array first. The old addJoinConstraint is now a wrapper for this new overload.
hborla
force-pushed
the
merge-joined-literal-typevars
branch
from
af01773 to
cd44ca8
Compare
|
@swift-ci please test |
|
Build failed |
|
Build failed |
|
@swift-ci please smoke test |
…ssage for each element whose type variable was merged in addJoinConstraint
|
@swift-ci please smoke test |
addJoinConstraintthat merges type variables for "atomic" literal expressions of the same kind. This is valid because such type variables will have the same set of constraints on them, and must be bound to the same type.addJoinConstraintto take in an iterator range and a callback to get the input types/locators. This allows callers to lazily compute the inputs-to-join rather than construct an array beforehand. The oldaddJoinConstraintis now a wrapper around this new overload.addJoinConstraintto join array element types.Resolves: rdar://problem/66176314
Resolves: rdar://problem/30596744