On 13.09.2007, at 21:19, Eric Covener wrote:
On 9/13/07, Erik Abele <[EMAIL PROTECTED]> wrote:
On 12.09.2007, at 16:12, Eric Covener wrote:
build.sh has:
# raise stack size...
ANT_OPTS="-Xmx128m -mx128m"
AFAICT those affect the heap only and to change the native thread
stack size java seems to want -ss/-Xss.
Yep, 'java -X' tells:
-Xms<size> set initial Java heap size
-Xmx<size> set maximum Java heap size
-Xss<size> set java thread stack size
Though the last time we had any mem-related problems it was
sufficient to simply raise the heap.
Adding -Xss10m allowed me to get past java.lang.StackOverflowError
during the docs build.
Hmm, considering the default thread stack size of 512kb on Sparc and
256kb on Intel for 1.3 and 1.4 32-bit VMs I think this is a bit too
much, no? (Since it quickly adds up to the whole amount of mem
used...)
Does it also work with 1mb, 2mb or 4mb?
Ah, and just out of interesst, what does 'ulimit -s' say for you? On
OS X it's 8192k so 10m is kind of moot here anyway, though shouldn't
do any harm :)
FWIW on this system (IBM java5 ppc32) 512k fails, 1m succeeds.
Since nobody else chimed in so far I've just committed a ppc-specific
fix for build.sh - IBMs Java 5 is known to cause several issues
there... (you may also want to try 1.4.2 which is more stable IIRC).
Cheers,
Erik
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
