Fixes for various txxxx tickets #86

odersky · 2014-03-20T12:57:00Z

No description provided.

odersky · 2014-03-20T12:58:18Z

This is a long list of fixes to make existing Scala test cases pass, or to invalidate the test cases. Review by @gzm0 and @sjrd, sheparding by @gzm0 please.

DarkDimius · 2014-03-20T13:00:52Z

Please not that first two commits are already effectively merged into master, but with different sha1(changed during rebase)

odersky · 2014-03-20T13:02:23Z

Yes, I had lots of merge conflicts when I pulled from master. Maybe that's because of this.

gzm0 · 2014-03-20T13:07:45Z

I will remove these two commits (they undo each other)

sjrd · 2014-03-20T15:08:39Z

That's all for me.

gzm0 · 2014-03-20T19:35:54Z

Otherwise LGTM for me. I will update the commits.

gzm0 · 2014-03-21T09:38:14Z

Updated commits according to comments. IMHO we can merge this as soon as CI passes.

sjrd · 2014-03-21T09:45:05Z

Yep, except that apparently GitHub can't merge it automatically ... Rebase on top of master?

gzm0 · 2014-03-21T09:55:55Z

Aarrrgh... Is there I way I can see that? Will rebase

Test t0288 moved to disabled due to lack of Java interop. Test t0273 fixed by relaxing double def condition: We only regard two definitions that define the same name and have exactly the same signature as double definitions. Previously, signatures that defined the same parameters were also excluded.

Two fixes were needed 1) When typing a function value (x1: T1, ..., xN: Tn) => e, don't unconditionally issue an error if the expected function type arity is different from N. Instead, issue an error only if one of the types T1, ..., Tn is absent. The idea is that only then we need to consult the expected type for the parameter type. This allows to fix the problem later by an implicit conversion applied to the function value. 2) When eta-expanding, do not automtically take the arity of the expected function value as the arity of the generated lambda. Instead, take the method's arity, and copy method parameters into the lambda in case the arities are different.

_root_ is now entered into an enclosing context.

A debug assertion in implicitSearch gave a false alarm and was removed.

Added more tests which all pass, except for tests in disabled and pending. t0694 went from pos to neg, because the kind of alias type used in t0695 is no longer supported.

Previously, only implicit method types were eligible as views. This is too strict, as it rules out view bounds. We now also consider types that derive from Function1. The reason for not allowing any type is that this would cause us to check many more types for applicability when an implicit view is searched.

Was previously wrapped in a package <empty> but the resulting tree had no position, which caused a Typer assertion. If now represented as EmptyTree.

which all pass.

Method type comparison via <:< yielded false if the signatures of the two method types differed. This is too strict, because methods can have the same parametyers but different result types and still be in a subtype relationship. We now onyl demand that the sighatures have the same parameters.

type T was not recorgnized as a SAM type because a case was missing in zeroParamClass.

Needed an extra case in isSubType.

Need to do unit discarding also in selection prototypes.

The original test is now in error because the type Meta in the prefix Meta#Event is not stable and contains an abstract member Slog. Even after removing Slog, the test in pos was still in error because the bound type parameters were incorrectly recognized as abstract members. This has been fixed by the changes to Types.

The type of self name "x" was taken to be the thisType of the current owner. But the current owner was a local dummy of the class in question, so the ThisType was NoPrefix. Inserting an enclosingClass fixes the problem.

(and also of t0625, which reappeared). Several fixes were made. In summary: 1. Naming and representation of KigherKinded traits changed. It's now $HigherKinded$NIP where the letters after the second $ indicate variance (N)egative, (I)nvariant, (P)ositive. The HKtraits themselves are always non-variant in their parameters. 2. When deriving refined types over higher-kinded types, the variance of a type alias is the variance of the new type constructor. 3. isSubTypeHK was changed, as was the position from where it is called. 4. appliedTo also works for PolyTypes.

Mostly Java interop tests which are not yet supported. The test infrastructure for Java ocmpilation and the java parser from Scala are still missing.

Turned parameter into receiver (reciever was not used before at all).

baseTypeWithArgs now also keeps track of refinements in the subtypes. Without that, the approximated lub in t1279a is too coarse and the program fails to typecheck.

test case t0273. Was positive in Scala 2, is now deemed to be negative. Two two definitions def a = () => () def a[T] = (p:A) => () do have matching signatures, so should constitute a double definition. I previously thought that we can get away if the two definitions have different result types, but then you immediately have a problem because the denotations have matching signatures for the pruposes of "&" yet cannot be merged. Which of the two definitions would override a definition in a base class is then an arbitrary decision.

stripImplicits needs to take polytypes into account.

gzm0 · 2014-03-21T10:28:56Z

Rebased.

sjrd · 2014-03-21T11:20:01Z

And now travis failed ...

gzm0 · 2014-03-21T12:02:47Z

@odersky this is caused by the rebase itself and is an incompatibility between the changes with #88.

The tips before my rebases are as follows:

gzm0/dotty@4d7ee1d Your original branch
gzm0/dotty@e96c43d Two initial commits removed
gzm0/dotty@0e6a79e Cleaned up according to comments
dotty-staging/dotty-1@5f318bc After rebase (this branch)

The diff between the tips before the rebase on master is trivial (gzm0/dotty@4d7ee1d ... gzm0/dotty@0e6a79e)

diff --git a/src/dotty/tools/dotc/core/TypeComparer.scala b/src/dotty/tools/dotc/core/TypeComparer.scala
index ba411d4..d66d406 100644
--- a/src/dotty/tools/dotc/core/TypeComparer.scala
+++ b/src/dotty/tools/dotc/core/TypeComparer.scala
@@ -921,10 +921,10 @@ class TypeComparer(initctx: Context) extends DotClass {

   /** Try to distribute `&` inside type, detect and handle conflicts */
   private def distributeAnd(tp1: Type, tp2: Type): Type = tp1 match {
+    // opportunistically merge same-named refinements
+    // this does not change anything semantically (i.e. merging or not merging
+    // gives =:= types), but it keeps the type smaller.
     case tp1: RefinedType =>
-      // opportunistically merge same-named refinements
-      // this does not change anything semantically (i.e. merging or not merging
-      // gives =:= types), but it keeps the type smaller.
       tp2 match {
         case tp2: RefinedType if tp1.refinedName == tp2.refinedName =>
           tp1.derivedRefinedType(
diff --git a/tests/neg/t0654.scala b/tests/neg/t0654.scala
index b4a3f68..52dbbb0 100644
--- a/tests/neg/t0654.scala
+++ b/tests/neg/t0654.scala
@@ -2,5 +2,4 @@ object Test {
   class Foo[T]
   type C[T] = Foo[_ <: T]   // error: parameter type T of type alias does not appear as type argument of the aliased class Foo
   val a: C[AnyRef] = new Foo[AnyRef] // follow-on error: wrong number of type arguments for Test.C, should be 0
-
 }
diff --git a/tests/pos/t1292.scala b/tests/pos/t1292.scala
index 2baa9d8..8e69734 100644
--- a/tests/pos/t1292.scala
+++ b/tests/pos/t1292.scala
@@ -9,8 +9,7 @@ trait Foo[T <: Foo[T, Enum], Enum <: Enumeration] {

 trait MegaFoo[T <: Foo[T, Enum], Enum <: Enumeration] extends Foo[T, Enum] {
   def doSomething(what: T, misc: StV, dog: Meta#Event) = None
-    // error: Meta is not a valid prefix for '#'.
-    // The error is correct. Meta is not stable, and it has an abstract type member Slog
+
   abstract class Event
   object Event

Fix of d6df293. It turned out the original commit was faulty in that iterator.flatten did not typecheck. The problem is fixed in this commit and flatten is added to the collections test.

odersky · 2014-03-21T17:40:53Z

Should be OK now. Over to @gzm0

gzm0 · 2014-03-24T16:27:06Z

LGTM @sjrd?

sjrd · 2014-03-24T16:48:49Z

LGTM

Fixes for various txxxx tickets

Backport "fix: don't consider `into` as a soft-modifier" to 3.3 LTS

odersky added 21 commits March 21, 2014 11:24

Fix of t0504: _root_ not found

4b57b3d

_root_ is now entered into an enclosing context.

Fix of t0591: implicitly

6f1ef32

A debug assertion in implicitSearch gave a false alarm and was removed.

More tests

6436fa6

Added more tests which all pass, except for tests in disabled and pending. t0694 went from pos to neg, because the kind of alias type used in t0695 is no longer supported.

Fix of t0774: empty file

7bf837c

Was previously wrapped in a package <empty> but the resulting tree had no position, which caused a Typer assertion. If now represented as EmptyTree.

More tests

90f430b

which all pass.

Fix of t1056 - PartialFunction

7820e62

type T was not recorgnized as a SAM type because a case was missing in zeroParamClass.

Fix of t112606A - path dependent types

0a1ec31

Needed an extra case in isSubType.

Fix of t1123 - Unit discarding

474b35f

Need to do unit discarding also in selection prototypes.

More tests

c854cc7

Fix of t1280 - type of self

40202ee

The type of self name "x" was taken to be the thisType of the current owner. But the current owner was a local dummy of the class in question, so the ThisType was NoPrefix. Inserting an enclosingClass fixes the problem.

More tests, currently disabled

0dda8a1

Mostly Java interop tests which are not yet supported. The test infrastructure for Java ocmpilation and the java parser from Scala are still missing.

Refactored lookupRefined

7e1343e

Turned parameter into receiver (reciever was not used before at all).

Fix of t1279a: baseTypeWithArgs

07939c9

baseTypeWithArgs now also keeps track of refinements in the subtypes. Without that, the approximated lub in t1279a is too coarse and the program fails to typecheck.

Fix of t1272: overloading resolution

5f318bc

stripImplicits needs to take polytypes into account.

Fix of Fix of t1236: higher-kinded

9b17c6c

Fix of d6df293. It turned out the original commit was faulty in that iterator.flatten did not typecheck. The problem is fixed in this commit and flatten is added to the collections test.

odersky assigned gzm0 Mar 21, 2014

sjrd added a commit that referenced this pull request Mar 24, 2014

Merge pull request #86 from dotty-staging/fixes

83449e1

Fixes for various txxxx tickets

sjrd merged commit 83449e1 into scala:master Mar 24, 2014

chrislailee mentioned this pull request Jun 18, 2021

Stack overflow in ImportSuggestions#distinctRefs #12876

Closed

WojciechMazur pushed a commit to WojciechMazur/dotty that referenced this pull request Mar 19, 2025

Merge pull request scala#86 from scala/backport-lts-3.3-21924

c684200

Backport "fix: don't consider `into` as a soft-modifier" to 3.3 LTS

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixes for various txxxx tickets #86

Fixes for various txxxx tickets #86

odersky commented Mar 20, 2014

odersky commented Mar 20, 2014

DarkDimius commented Mar 20, 2014

odersky commented Mar 20, 2014

gzm0 commented Mar 20, 2014

sjrd commented Mar 20, 2014

gzm0 commented Mar 20, 2014

gzm0 commented Mar 21, 2014

sjrd commented Mar 21, 2014

gzm0 commented Mar 21, 2014

gzm0 commented Mar 21, 2014

sjrd commented Mar 21, 2014

gzm0 commented Mar 21, 2014

odersky commented Mar 21, 2014

gzm0 commented Mar 24, 2014

sjrd commented Mar 24, 2014

Fixes for various txxxx tickets #86

Fixes for various txxxx tickets #86

Conversation

odersky commented Mar 20, 2014

odersky commented Mar 20, 2014

DarkDimius commented Mar 20, 2014

odersky commented Mar 20, 2014

gzm0 commented Mar 20, 2014

sjrd commented Mar 20, 2014

gzm0 commented Mar 20, 2014

gzm0 commented Mar 21, 2014

sjrd commented Mar 21, 2014

gzm0 commented Mar 21, 2014

gzm0 commented Mar 21, 2014

sjrd commented Mar 21, 2014

gzm0 commented Mar 21, 2014

odersky commented Mar 21, 2014

gzm0 commented Mar 24, 2014

sjrd commented Mar 24, 2014