When I try the simple code:

use strict;
use warnings;


while (<DATA>) {
    s/://g;
    print $_;
}

__DATA__
9:23:00
02:23:34
[download]

Unless you're using the -i command line option, you can't edit the file in place. You need to open the original file for input, open a temporary file for output, read the orig, get rid of the colons, write the results to the temp file, and then move the temp file to replace the orig.

That ought to do the trick.

Of course if you're just trying to remove colons from times, and preserve them elsewhere, this solution is too aggressive. But your original question didn't really specify.

sed s/://g file.txt > newfile.txt
[download]

Recent versions of GNU sed have stolen Perl's -i option, so you can say

sed -i 's/://g' file.txt
[download]

and have file.txt updated in-place.

Makeshifts last the longest.

One of many many ways, from the command line perl -pi -e's/\b(\d\d):(\d\d):\d\d\b/$1$2/g;' filename.ext \b is a word boundary, which helps narrow the selection. \d are digits, and we capture the significant ones with parens. That makes them available as $1, $2 in the substitution string. The /g modifier makes the substitution for all instances on a line.