Several simplifications to TypeSymbolWithAnnotations #28583
Conversation
| @@ -138,21 +138,29 @@ internal TypeSymbolWithAnnotations SubstituteType(TypeSymbolWithAnnotations prev | |||
| /// In particular, if substitution makes type tuple compatible, transform it into a tuple type. | |||
| /// </summary> | |||
| internal TypeSymbolWithAnnotations SubstituteTypeWithTupleUnification(TypeSymbol previous) | |||
There was a problem hiding this comment.
More general question: why do we have this API, instead of just using SubstituteType?
There was a problem hiding this comment.
Because the compiler does not match the language in the way that tuple types are handled. See #20648 which describes a set of changes that are planned to bring them into alignment.
There was a problem hiding this comment.
(I'm afraid I may have answered the question you didn't ask)
In reply to: 202755152 [](ancestors = 202755152)
| @@ -304,15 +327,20 @@ public bool IsAtLeastAsVisibleAs(Symbol sym, ref HashSet<DiagnosticInfo> useSite | |||
| return NullableUnderlyingTypeOrSelf.IsAtLeastAsVisibleAs(sym, ref useSiteDiagnostics); | |||
| } | |||
|
|
|||
| public virtual TypeSymbolWithAnnotations SubstituteType(AbstractTypeMap typeMap) | |||
There was a problem hiding this comment.
virtual [](start = 15, length = 7)
What's the virtual removal change related to?
There was a problem hiding this comment.
The new protected overload is now virtual so derived classes have a single method to override rather than two.
|
Done review pass (commit 2) |
jcouv
added
Area-Compilers
New Language Feature - Nullable Reference Types
Minor changes to
TypeSymbolWithAnnotations:ToDisplayStringto base classwithTupleUnificationparameter toSubstituteTypeLazyNullableTypeParameter.SpecialTypeandIsRestrictedTypeCode changes separated out from #28453.