Closed scabug closed 6 years ago
Imported From: https://issues.scala-lang.org/browse/SI-4512?orig=1 Reporter: @odersky
@paulp said: (In r24964) Temporarily sending lisp.scala to be interpreter.scala's equally memory hungry buddy in pending. References #4512. This should be straightened out imminently, but I will solve this more easily by looking forward, not backward. No review.
Commit Message Bot (anonymous) said: (extempore in r25047) Restored the lisp.scala test. I already restored the intepreter test in r25038, so closes #4512, no review.
@SethTisue said: reopening because the interpreter test was restored but the lisp test is still disabled and the issue with both tests appears to be the same
@SethTisue said: see http://groups.google.com/group/scala-internals/browse_frm/thread/cbba8367ce26ba28 for fresh analysis on this
@SethTisue said: partest defaults tweaked, lisp test reinstated by Paul in r25387. hopefully it won't start failing the build again. optimistically closing the ticket.
@SethTisue said: should this problem come back, Jason Zaugg suggests "add -XX:+HeapDumpOnOutOfMemoryError to the JVM options and archive the heap dumps as build artifact for forensics"
@paulp said: Lots of failures again implicating the lisp test. Collecting heap dump impractical for me. Will take pointers.
@adriaanm said (edited on Jun 11, 2012 1:03:07 PM UTC):
(be warned, that goal is not just a git mv
away)
another dead Lisp dream.
I moved test/files/jvm/interpreter.scala to pending for now. It stops running on more and more machines because default heap size is not enough for it. It seems that my addition of 55 bridge forwarder methods and 5 new interfaces pushed it over the edge here. We should find pass a sufficient heapsize to the subprocess that runs the interpreter and then bring the test back.