Fix #5279: Be more careful where we create symbolic refs from named ones #5287
Conversation
odersky
force-pushed
the
fix-#5279
branch
3 times, most recently
from
6ec2a91 to
7cf7582
Compare
Be more where we create symbolic refs from named ones.
Leave name designators as they are. It makes a difference for the info of the reference.
|
test performance please |
|
performance test scheduled: 2 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/5287/ to see the changes. Benchmarks is based on merging with master (d5b885d) |
|
This is now part of #5300. Anyone care to review it independently? |
There was a problem hiding this comment.
The following code fails to compile:
trait BB {
val a: Any
}
trait TT { self: { type T = Int } & BB { val a: T } =>
type T
def lift(x: Int): T = a
}Error message:
-- [E007] Type Mismatch Error: tests/pos/i5279.scala:7:24 ----------------------
7 | def lift(x: Int): T = a
| ^
| found: TT.this.T(TT.this.a)
| required: TT.this.T
|
one error found| /** Hook that can be called from creation methods in TermRef and TypeRef */ | ||
| def validated(implicit ctx: Context): this.type = { | ||
| this | ||
| } |
There was a problem hiding this comment.
Yes, I do use if occasionally for debugging, and decided to leave it in for others to use as well if needed.
| adapted | ||
| else this | ||
| result.setDenot(denot) | ||
| result.asInstanceOf[ThisType] |
There was a problem hiding this comment.
I guess adapted.eq(this) is a performance tweak, semantically we can drop it without changing the result.
This looks like another instance of the double vision problem described in #5300. |
No description provided.