At 23:34 -0700 2006/05/30, Joey K Tuttle wrote:
There is considerable improvement in j601 on this issue, but I'm still
surprised at the seemingly small limit.
Now that I have a working j601 in a Linux environment, I ran
some of the same tests and found that the limit on !# arguments
is more severe in Linux than in Darwin. Plus there seems to be
a more serious bug that crops up ...
[EMAIL PROTECTED]:~/test$ ./mkfiles 100
j601/beta/2006-05-20/22:30
Running in: Linux
[EMAIL PROTECTED]:~/test$ ./atest 1*
102 7040
This looks OK, 7 Kbytes and the 100 files + first two items in ARGV
[EMAIL PROTECTED]:~/test$ ./mkfiles 200
j601/beta/2006-05-20/22:30
Running in: Linux
[EMAIL PROTECTED]:~/test$ ./atest 1*
148 10496
Ooops!! Something is wrong here. The argument count should be
202, not 148 -- this seems to be a bug (and would be serious
in a real application)
[EMAIL PROTECTED]:~/test$ ./mkfiles 250
j601/beta/2006-05-20/22:30
Running in: Linux
[EMAIL PROTECTED]:~/test$ ./atest 1*
148 10496
[EMAIL PROTECTED]:~/test$ ls 1* | wc
250 250 1250
More files - but the argument count and size is the same...
[EMAIL PROTECTED]:~/test$ ./mkfiles 280
j601/beta/2006-05-20/22:30
Running in: Linux
[EMAIL PROTECTED]:~/test$ ./atest 1*
Segmentation fault
[EMAIL PROTECTED]:~/test$
Finally we hit the wall that crashes j, but this is at a much
lower point than in Darwin - and I would really like to see the
number of arguments to a !# script be > 1000 (or higher).
- joey
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
