print problem

[Punto]

Hi..

I have a loop that does a sleep(1); and then a number of calculations, and then I display the results like this:

print "$result1 - $result2 - $result3\r";

I'm using \r because I want it to be overwritten on the next print, but it doesn't print anything.. if I change to \n, it works fine.. Do I have to flush a buffer or something?

Thanks..

[{NULE}]

Hi Punto,

Yes, is the short answer to your question. Here is some code that works:

#! /usr/bin/perl -w
use strict;

$| = 1; # This forces your selected filehandle (STDOUT
        # in this case) to be unbuffered.
my $i = 0;
while (1)
{
        sleep(1);
        print $i++," - ",$i++," - ",$i++,"\r";
}

There are other ways to do this that might be more elegant or more cross-platform. I believe that looking on CPAN for Term and Curses might be useful to you. (Note to self: finish up that stinking Curses tutorial already...)

Good luck,
{NULE}
--
http://www.nule.org

[hossman]

The reason you need to set $| is explained in "perldoc perlvar"...

    $|    If set to nonzero, forces a flush right away and
          after every write or print on the currently
          selected output channel.  Default is 0
          ...
          STDOUT will typically be line buffered if output
          is to the terminal and block buffered otherwise.
          ...

So in your orriginal program, the buffer was just building up because it never say a line termination. (and when you switched to "\n" it did).

[mr_mischief]

The most accurate short answer is maybe.

On Linux Mandrake 8.0 through PuTTY and on Windows 98 SE with ActiveState build 623 I didn't need to flush buffers doing something similar. Different term types on different systems will do different things with the code I used. Considering the problem you are having, I'd say flushing buffers is a good idea. The $| variable is probably your very best bet, although the Term modules, Curses module, or possibly a few other things could help. I won't mention the other things, because using the three I already mentioned should be more than enough to get the job done and are widely used. There are, in the Perl spirit, even more ways to do it, but you probably will never need to use them. ;-)

Just in case you're curious, here's the Linux test code I used:

perl -e 'for (a..fred) { print "$_\r" } print "\n"'

...and here's the Windows 98 code:

perl -e "for (a..f) { sleep 1; print qq{$_\r} } print qq{\n};"

or

perl -e "for (a..fred) { print qq{$_\r} } print qq{\n};"

[Foo::Bar]

All good solutions but there is another way to do it...

use FileHandle;
STDOUT->autoflush(1);

Should solve the problem also...

[dmmiller2k]

I won't pick nits about whether this makes STDOUT unbuffered, or merely behave as though it were unbuffered (yes, there is a difference), but in the spirit of T M T O W T D I:

$|=1;

is exactly equivalent to:

use English;
$OUTPUT_AUTOFLUSH=1;

which is effectively the same as:

use FileHandle;
STDOUT->autoflush(1);

It all boils down to how much typing you're up for vs. whether or not you'll remember what it means later.

dmm

You can give a man a fish and feed him for a day ...
Or, you can teach him to fish and feed him for a lifetime

[Anonymous Monk]

except that (unless you are going to use them elsewhere)

use English;  ...and...
use FileHandle;

both drag in a pile more (mostly useless) code... isn't

$| = 1; #set autoflush on

good enough?

[chromatic]

English isn't that bad. It could be made a little skimpier, though. Maybe I'll fix up its import slightly. Thanks for the idea!

[k2]

Hello! I do almost the same thing, in my case it's at "status bar"
printing how many lines my script has read from a file.
In my case I start by printing the "\r". Otherwise
you print you results and the emediatly erase them.
So without having tested this I think
 print "\r$result1 - $result2 - $result3";
will work.
/k2