Allow generic function instantiation of an existing function object #1812
Conversation
…on on an existing function object
eernstg
changed the title
|
@lrhn, I removed the review request because I'm revising this PR. I'll restore it when the text is stable again. |
|
The rules are now much more compositional (identical/== for the whole is determined by identical/== on the parts). Restored the review request. |
…on on the `call` of a function object
|
Added one change which was omitted by accident at first: The PR now also removes the rule which makes it an error to perform a generic method instantiation on the |
|
A few experiments with the current implementations show that we do actually allow for the tearoff. One might think "Sure, why not?", but we specified that it isn't possible, so I created #1827 to raise the question and perhaps decide that we should just support both the closurization and the generic function instantiation on an |
…ions This allows local variables and other expressions, as long as the static type is a generic function type, to be explicitly type instantiated. * Expressions with a non-generic function type cannot be type instantiated. * Expressions with a type of `Function` cannot be type instantiated. * Expressions with other non-function types cannot be type instantiated. This results in doubling-up of errors in a few situations. I think these are generally rare occurrences, and tricky to prevent double reporting, so I've left them for now. Bug: #46020, Change-Id: Iad212fd95773f39f3202480b3fa71f6a28c7698f dart-lang/language#1812 Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/211941 Commit-Queue: Samuel Rawlins <srawlins@google.com> Reviewed-by: Brian Wilkerson <brianwilkerson@google.com> Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
dart-lang/language#1812 Closes #47147 Change-Id: Iaba365a583bca456c19a6edba8801bab630304c5 Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/212590 Reviewed-by: Dmitry Stefantsov <dmitryas@google.com> Commit-Queue: Johnni Winther <johnniwinther@google.com>
Update section "Generic Function Instantiation" to allow this operation on an existing function object, in addition to the currently supported cases (where the expression denotes a declaration).
This update does not yet mention explicit instantiation, but I think the work needed to allow the current generalization and the work needed to support explicit instantiation are rather well separated. So it's not a waste of time to those two things in two steps (and we can then do 'constructor-tearoffs' as one single update, which is probably also nice).