seekable stdin

[Gyepi SAM]

On Thu, Dec 09, 2004 at 02:33:34PM -0700, Michael Gale wrote:
> I would rather not use the Perl one as a file is executed for each piece 
> of mail ... so perl would constantly have to be started ... if it was a 
> daemon that would be ok ...

This line of thinking comes up ever so often, but really needs to be examined.
On my system, /bin/sh is a link to /bin/bash, which is 58K in size, versus 14K
for perl. Perhaps you should write your filter in perl ;)
Honestly, if you were looking for optimizations, I would look at file IO and
data copying.

> How would the C program you have work ? Would I just have to execute it 
> via the script ??

The C program is attached. Just compile it and put it in a reasonable
location.  Then invoke it from your shell script.
Note that it will complain and exit with a non-zero code if
stdin is not seekable.

Just for kicks, I compared the perl and C versions. As expected the C version
is faster; by a factor of 4. But an extra 2.4 seconds per 1000 emails is not a
big deal either.

[gyepi at nome gyepi]$ time sh -c 'for i in `seq 1 1000`;do src/misc/rewind-stdin
< /dev/null;done'

real    0m0.515s
user    0m0.197s
sys     0m0.318s
[gyepi at nome gyepi]$ time sh -c 'for i in `seq 1 1000`;do perl -e "seek
STDIN,0,0" </dev/null;done'

real    0m2.366s
user    0m1.346s
sys     0m1.019s


-Gyepi
-------------- next part --------------
#include <stdio.h>
#include <errno.h>

int main(int argc, char *argv[])
{
  if (fseek(stdin, 0, 0) == - 1){
    perror(argv[0]);
    exit(1);
  }
  exit(0);
}

/*
Local Variables
compile-command: gcc -o rewind-stdin rewind-stdin.c
End:
*/