Final definite assignment #1091
Conversation
| @@ -433,7 +468,7 @@ Definitions: | |||
| - Else if `S` is `X extends R` then let `T1` = `X & T` | |||
| - Else If `S` is `X & R` then let `T1` = `X & T` | |||
| - Else `x` is not promotable (shouldn't happen since we checked above) | |||
| - Let `VM2 = VariableModel(declared, T1::promoted, T1::tested, assigned, | |||
| - Let `VM2 = VariableModel(declared, T1::promoted, S::tested, assigned, | |||
There was a problem hiding this comment.
Is this a deliberate change? It seems wrong--we want to add T1 to the "tested" set, since it's (effectively) the type we're testing against. S is the old type, so adding it to "tested" doesn't gain us anything.
There was a problem hiding this comment.
Yes, S was wrong. It's not clear to me that T1 is right though, is it? Or should it be T?
There was a problem hiding this comment.
Good point. The current implementation uses T, so let's change it to that. I'm happy to do that as a separate PR if you want (since at this point it's unrelated to this change), or you can go ahead and do it now--your call.
There was a problem hiding this comment.
Since I backed out the rest of the changes to this doc, how about you add this change to one of your updates?
|
Removed initialization based promotion, restored |
|
Tests out for review here: https://dart-review.googlesource.com/c/sdk/+/157881 |
New tests for the definite assignment related changes specified in dart-lang/language#1091 . Change-Id: I119e36c1bddca97f7021c9015b0f2eee5cb80969 Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/157881 Reviewed-by: Erik Ernst <[email protected]> Reviewed-by: Paul Berry <[email protected]>
|
Tests for the initializer promotion related behavior out for review here: https://dart-review.googlesource.com/c/sdk/+/163663 |
Final variant/proposal for definite assignment based inference.
This variant requires explicit initialization on all branches, but allows adjustment for Null.