Test::Benchmark ??

[Jim Cromie]

putting a name to it is painting a bullseye on it ;-)

I recently wrote a test to do a benchmark / performance test,
and found it somewhat difficult to get the output out to screen,
the STDOUT, STDERR takeover done by Test::More was
catching, and counting it as errors. (i think, but its not pertinent)
I ended up doing this..

for $data ($AR, $HR, [EMAIL PROTECTED], $ezpr, $ddo) {

   $rows = cmpthese(-3, {
       'DD'   => sub { Dumper ($data)},
       'EzDD' => sub { $ezdd->($data)},
   }, 'none');
  
   # kinda hacked way of getting output..
   for $r (@$rows) {
       diag sprintf("%12s %12s %12s %12s", @$r);
   }
}

IOW,  since Benchmark::cmpthese() printing was blocked, I suppressed it,
then printed the rows myself with diag().
SOOOO

Is there a Test::* enhancement latent here ?

I could hack Benchmark to accept print-styles "#$foo",
where $foo is :
     STYLE can be any of 'all', 'none', 'noc', 'nop'
                or 'auto'. 'all' shows each of the 5 times
thus sneaking in a prefix that might allow it out of the Test::More 
stdout/err box.
I dont know if this would work, but it sounds like a bad idea anyway-
it creates a dependency on a new Benchmark hack, which is in CORE,
and would then have to become dual-life.

Alternatively - Test::Benchmark could override the appropriate pieces
of Benchmark.  This is only worthwhile if print "# foo" actually gets 
thru the
test framework.

Finally, and perhaps most practically,  diag_code (&;$)  could run the
subroutine within a box that lets the prints thru.  The biggest problem 
is that
its probably hard to also catch errors/warnings etc that should be caught.

So what think yall ?