Igor.Vatolkin:
>> Hello Marco,
>>
>> I use the 0.8 version of JavaCompiler. I tried to minimize the example=
>> that leads to exception. Starting "ReadLines.exe fftshorttmp2.dat 16"
>> worked but "ReadLines.exe fftshorttmp2.dat 17" not any more.=20
>> But after I copied the files from my network drive to C: the program=20
>> worked without exceptions!! However it didn't work on C: with another =
=2Edat and the
>> number of lines about 2000 (as described in my first email).=20
>> One aspect which could probably explain the first test is that the net=
work
>> connection from my pc is slowly. But why doesn't it work from C: with
>> greater line number then? I use Java 1.4 and Windows XP SP2.=20
>> If I start the ReadLines from Eclipse (under Linux) no problems occur.=
=20
>> It seems for me that Windows may read the data too slow on my pc, but =
then I would
>> expect not "NullPointerException" but "OutOfTime" or anything else..
>> Maybe you can try to start ReadLines.exe once with the=20
>> attached file and tell me if it works fine on your system?
It runs fine(*) on my PC, the NullPounterException only appears if I try =
to read more lines than the file has.
(*) But there is something strange going on here, if you use Windows Task=
Manager to monitor CPU and memory used there is a big change between rea=
ding 16 and reading 17 lines, both the CPU and memory reading jump. Also=
between the first and subsequent runs, the executable is 2.4M (and I hav=
e 1G of RAM) so this is also strange.
I tried to test with a larger file, 2500 lines with the same second value=
, and the program flies through that file with no big CPU & memory jump. =
So my guess is that the java tokenizer is the one causing strange behavi=
our (at first I thought it was the file size, or the DOS terminators CR-L=
F, but that was not it).
I'm not sure where the problem (causing the jump in memory used) is; but =
it's not the java compiler, must be a library.
--=20
Ren=E9 Berber
|
