Am 2015-11-19 um 17:17 schrieb Marcus Lagergren:
Perhaps set a larger heap limit in the test if someone runs with a pathologic
JVM config, but no biggie.
I don't worry so much about the JVM config but the the actual hardware
limits of machines the test may run on. And memory depletes even quicker
if we @fork to set the max heap size, so I'd like to keep it that way.
Thanks for reviewing!
Hannes
+1
On 19 Nov 2015, at 14:44, Hannes Wallnoefer <hannes.wallnoe...@oracle.com>
wrote:
Please review JDK-8143304: Random failures when script size exceeds token
limits:
http://cr.openjdk.java.net/~hannesw/8059934/webrev/
This assigns 28 bits to both the position and the length in the parser token,
and makes sure source files don't exceed this limit in the parser.
Previously we used 32 bits for the position and only 24 for the length. The
thinking behind that probably was that a token would never be as long as the
whole script but that turns out to be wrong (like for the main script function,
or a very long object or array literal).
Thanks,
Hannes
