[4.1][CSBindings] Form bindings correctly when they come from 'OptionalObject' constraint #14627
Conversation
…ect' constraint Fixes a bug in `getPotentialBindings` when the source of the bindings is 'OptionalObject' constraint and type variable is on the left-hand side of that constraint, that makes such type variable always have an optional type since right-hand side of 'OptionalObject' is its 'object' type. Resolves: rdar://problem/37508855 (cherry picked from commit 5208049)
|
@swift-ci please test |
|
@swift-ci please test source compatibility |
|
Build failed |
|
@rudkx Looks like the test |
|
@swift-ci please test macOS platform |
|
@xedin Okay - do you want to disable in the 4.1 branch and open a JIRA and I will take a look at it on master? We may want to move away from running the timed tests in CI. They might be more trouble than they are worth at the moment. |
|
@rudkx I think it would be fine if you increase the threshold to 1.5 or something, it seems like it's only happening on one of the platforms. |
|
@rudkx I'll try to do that a bit later today if you don't to that before me. |
|
I opened a JIRA as well as PRs to disable it everywhere for now. |
|
@rudkx Thanks! |
|
@swift-ci please nominate |
• Explanation: Fixes a bug in
getPotentialBindingswhen the source of the bindingsis 'OptionalObject' constraint and type variable is on the left-hand
side of that constraint, that makes such type variable always have an
optional type since right-hand side of 'OptionalObject' is its 'object'
type.
• Scope of Issue: Affects logic related to picking bindings for type variables in constraint solver.
• Risk: Low risk; Fixes a bug in constraint solver.
• Reviewed By: @rudkx
• Testing: Compiler regression tests
• Radar / SR: rdar://problem/37508855
Resolves: rdar://problem/37508855
(cherry picked from commit 5208049)