Preliminary constraint solver support for variadic generics #61465
Conversation
slavapestov
force-pushed
the
variadic-generics-constraints
branch
from
4d747da to
bdc8169
Compare
slavapestov
force-pushed
the
variadic-generics-constraints
branch
2 times, most recently
from
5d8f3a2 to
4476813
Compare
|
@swift-ci Please smoke test |
The expression type checker replaces type variables with UnresolvedType and PlaceholderType sometimes, so allow them to appear here.
…n unlabeled element
I'm also going to use it in TypeMatcher, for the Requirement Machine's same-type requirement desugaring.
…ns with new utilities
…nsions into account
slavapestov
force-pushed
the
variadic-generics-constraints
branch
from
4476813 to
b9d7201
Compare
|
@swift-ci Please smoke test |
| auto *packType = cast<PackType>(desugar2); | ||
|
|
||
| if (packExpansionType->getPatternType()->is<TypeVariableType>()) | ||
| return matchTypes(packExpansionType->getPatternType(), packType, kind, subflags, locator); |
There was a problem hiding this comment.
Is the idea here is that we'd like to resolve a pattern type of expansion using a pack type (type2) but at the same we don't want to double-check the matching if the pattern is already resolved? Could there be a situation when either pattern or pack type are only partially resolved and need matching to facilitate further inference?
There was a problem hiding this comment.
It was a hack to work around other problems in PackExpansionMatcher.cpp. This chunk of code was removed in subsequent PRs.
There was a problem hiding this comment.
Sounds good! I'll working through the list :)
|
|
||
| bool TupleType::containsPackExpansionType() const { | ||
| for (auto elt : getElements()) { | ||
| if (elt.getType()->is<PackExpansionType>()) |
There was a problem hiding this comment.
Why not make this a property of a tuple type?
There was a problem hiding this comment.
That's a good idea. I wanted to wait to micro-optimize until more code is fleshed out though.
No description provided.