diff options
author | Adriaan Moors <adriaan.moors@typesafe.com> | 2016-07-17 22:04:46 -0700 |
---|---|---|
committer | Adriaan Moors <adriaan@lightbend.com> | 2016-08-11 10:59:17 -0700 |
commit | 3540ffc1fb81eef75aeff41a0ba9142b1cce8a53 (patch) | |
tree | 45586d44ba0bc4914143e9412896c01ec1d61c52 /test | |
parent | 4555681544ec40c5b787b9708266fcbdcc2ff62b (diff) | |
download | scala-3540ffc1fb81eef75aeff41a0ba9142b1cce8a53.tar.gz scala-3540ffc1fb81eef75aeff41a0ba9142b1cce8a53.tar.bz2 scala-3540ffc1fb81eef75aeff41a0ba9142b1cce8a53.zip |
Admit @volatile on accessor in trait
There's no other place to squirrel away the annotation
until we create a field in a subclass.
The test documents the idea, but does not capture the
regression seen in the wild, as explained in a comment.
Diffstat (limited to 'test')
-rw-r--r-- | test/files/pos/trait_fields_volatile.scala | 13 |
1 files changed, 13 insertions, 0 deletions
diff --git a/test/files/pos/trait_fields_volatile.scala b/test/files/pos/trait_fields_volatile.scala new file mode 100644 index 0000000000..030b24f187 --- /dev/null +++ b/test/files/pos/trait_fields_volatile.scala @@ -0,0 +1,13 @@ +// This test illustrates the intent of what should work (but didn't for a while during the fields refactoring), +// but it does not actually defend against the regression seen in twitter-util's Scheduler, which I cannot reproduce +// outside the project. The whole project consistently fails to build before, and compiles after the commit +// that includes this test, but this single test file (as well as Scheduler.scala with external dependencies removed) +// compiles both before and after.... +// (https://github.com/twitter/util/blob/6398a56923/util-core/src/main/scala/com/twitter/concurrent/Scheduler.scala#L260-L265) +// There's also a run test that checks that the field in C is actually volatile. +trait Vola { + @volatile private[this] var _vola = "tile" + @volatile var vola = "tile" +} + +class C extends Vola |