diff options
author | Jason Zaugg <jzaugg@gmail.com> | 2013-01-08 18:36:30 +0100 |
---|---|---|
committer | Jason Zaugg <jzaugg@gmail.com> | 2013-01-26 15:25:24 +0100 |
commit | 982633a77ccef39e2c2611b88294d39d022a0f03 (patch) | |
tree | bc009ca501568457072db2a2b065fdd41c6682db /test/pending/pos/t4790.scala | |
parent | f01af109ae975461fe5a3120a69814521968fcce (diff) | |
download | scala-982633a77ccef39e2c2611b88294d39d022a0f03.tar.gz scala-982633a77ccef39e2c2611b88294d39d022a0f03.tar.bz2 scala-982633a77ccef39e2c2611b88294d39d022a0f03.zip |
SI-6556 Remove unneeded workaround in erasure.
In fb6e687, a workaround was added, described as:
> See SI-6556. It seems in some cases the result constructor
> type of an anonymous class is a different version of the class.
> This has nothing to do with value classes per se.
> We simply used a less discriminating transform before, that
> did not look at the cases in detail.
> It seems there is a deeper problem here, which needs
> following up to. But we will not risk regressions
> in 2.10 because of it.
I was expecting an arduous investigation into this,
but happilly the underlying problem was already solved
in the fix for SI-6648 (1587a77e). (Chalk up another
win for libscala's scala-hash!)
This commit escalates the logging to an assertion.
It also reverts the other change in fb6e687,
which introduced a case for RefinedType, which are
not needed to pass pos/t6556.scala.
Diffstat (limited to 'test/pending/pos/t4790.scala')
0 files changed, 0 insertions, 0 deletions