On 27/01/12 09:55, Damian Steer wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 27/01/12 09:37, Andy Seaborne wrote:
On 26/01/12 22:54, Damian Steer wrote:
The datasets were all being retained. Why this affected the mac
and not linux I have no idea.
Quite weird. It does not really explain the observations as a large
heap should at least move the point of failure around a bit.
See "Maven forks the tests ... and doesn't pass on options like dump
heap on oom.". I assume it also wasn't passing on Rob's heap increase.
(I wonder if the mac issue concerns the process launching?)
I see don't see why the mac is different though. It's the heap that's
running out. If the test object is being held onto, so dataset is
reachable and not a GC candidate, then it is so on all systems.
As it works on Oracle preview of jdk 7, we can run and hide behind
saying its a JVM bug or JVM/OS interaction. Weak!
This is the scripted tests - that might be a factor.
It will remain a mystery. Until next time.
Rob - it should work now.
Andy
Damian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAk8idH0ACgkQAyLCB+mTtyk1uACgvmurFSkEYQzUrmNWXKOvj3bQ
o3cAnR4+3hihoYBfpxAv1RToIcxE8KDk
=di1D
-----END PGP SIGNATURE-----
