Remove anomalies and gaps in handling annotations #13348
Conversation
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit https://dotty-bench.epfl.ch/13348/ to see the changes. Benchmarks is based on merging with master (68044a6) |
|
test performance please |
1 similar comment
|
test performance please |
odersky
changed the title
odersky
changed the title
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit https://dotty-bench.epfl.ch/13348/ to see the changes. Benchmarks is based on merging with master (6a3cd1f) |
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit https://dotty-bench.epfl.ch/13348/ to see the changes. Benchmarks is based on merging with master (f7344ab) |
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 0 running. |
|
Performance test finished successfully: Visit https://dotty-bench.epfl.ch/13348/ to see the changes. Benchmarks is based on merging with master (a82a1a6) |
Always print annotation arguments if there are some, except for Body annotations
A TypeMap previously mapped the annotation only if it returned a different result for the parent type. There is no good reason for this behavior. We now map always, but allow the mapping operation to be defined in the annotation. The change uncovers a bug illustrated in annotDepMethType.scala, which is fixed in the next commit.
Fix problems handling function types that are dependent through their result annotations.
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 0 running. |
|
Performance test finished successfully: Visit https://dotty-bench.epfl.ch/13348/ to see the changes. Benchmarks is based on merging with master (a82a1a6) |
| @@ -607,7 +607,7 @@ class RefinedPrinter(_ctx: Context) extends PlainPrinter(_ctx) { | |||
| case tree: Template => | |||
| toTextTemplate(tree) | |||
| case Annotated(arg, annot) => | |||
| toTextLocal(arg) ~~ annotText(annot) | |||
| toTextLocal(arg) ~~ toText(annot) | |||
There was a problem hiding this comment.
I think here we are now printing annot as a tree, and not a tree interpreted as an annotation
| class Foo() extends Annotation | ||
| class Bar(s: String) extends Annotation | ||
|
|
||
| def x: Int @nowarn @main @Foo @Bar("hello") = "abc" // error |
There was a problem hiding this comment.
with these changes -Xprint:parser will only show <none> where the class name should be
def x: Int @<none> @<none> @<none> @<none>("hello") = "abc"There was a problem hiding this comment.
in 3.1.0-RC1 we still show the correct class names with -Xprint:parser
There was a problem hiding this comment.
Good catch! Is there an easy way to write a printing test to guard against regressions for this?
There was a problem hiding this comment.
the PrintingTest can be modified to add cases that print at parser instead:
compiler/test/dotty/tools/dotc/printing/PrintingTest.scala
There was a problem hiding this comment.
@odersky I have added tests to PrintingTest in latest commit
Three main fixes, with one commit for each: