Nicolas Schier wrote:
>-while (<>) {
>+while (<STDIN>) {
That gets rid of all the dangerous and bizarre behaviour. It is `safe'
in that sense. However, it would leave the program still accepting
extraneous arguments (and now ignoring them). If I were patching the
program I'd want to also change it to signal a usage error when extraneous
arguments are supplied. (Input checking and good error messages are
useful features. Accepting extra arguments and ignoring them is a
recipe for users to develop spurious beliefs that extra arguments will
do something useful.) This pseudo-patch would add such argument checking:
-) || die "usage: ts [-r] [-i | -s] [-m] [format]\n";
+) && @ARGV <= 1 or die "usage: ts [-r] [-i | -s] [-m] [format]\n";
If you have the argument checking then the patch to change <> to <STDIN>
is not strictly necessary, because @ARGV would then always be empty by the
time that loop is reached. But it would still be good style to change <>
to <STDIN>, to better indicate the program's intended behaviour.
>If it is, I would like to include your reasoning into a patch for
>upstream, ok?
Sure. You may use that text freely.
-zefram
