HI Shawn.
Your email prompted me to look more carefully at my ps output. I do see
a few other java items in the output, and one of them is this:
noaccess 750 1 0 Nov 15 ? 3:10 /usr/java/bin/java
-server -Xmx128m -XX:+BackgroundCompilation -XX:PermSize=32m
When I shrunk the sleep time in my test program to zero and ran it in a
loop, a truss of the above process didn't react to it, so I'm not sure
the above process is connected in any way to my test program...
Supposing it was connected just for the heck of it...
Pmap for that command did show up quite a few shared regions, but pmap
for the footprint test program showed only a few shared regions:
Kbytes RSS Anon
B76F8000 1520 1520 - - r--s- rt.jar
FE420000 40 40 - - r--s- jhall.jar
FE432000 40 40 - - r--s- gnome-java-bridge.jar
FE590000 40 40 - - r--s- jhall.jar
FE5A0000 32 32 - - rw-s- 5578
FE6F8000 8 8 - - rwxs- [ anon ]
Total possibly shared RSS adds up to 1680Kb, still much less than 25Mb.
So, even if it is shared, it doesn't look like it makes that much of a
difference.
Thanks,
Jack
Shawn Walker wrote:
> On 16/11/2007, Jack Schwartz (Home Login) <Jack.A.Schwartz at sun.com> wrote:
>> Hi everyone.
>>
>> Yesterday at the Distro-Constructor meeting, I took an AI to look into
>> footprints of Java vs Python.
>
> Isn't a java vm shared while a python interpreter is not?
>
