Re: Which way is recommended ?

[betterworld]

instead of directly using $_

The term "directly" is not quite correct here, because it suggests that $_ is used in any case.

$_ will not be set if you write while( my $line = <$lfh> ).

In my opinion, using $_ is not really elegant. It's a global variable and it will be visible to called subroutines even if you localize it. Well, in perl 5.10 you can avoid this my using my $_. But unlike with named variables, you won't get any errors from strict if you forget to my the variable.

[DStaal]

Also, a good part of Perl's operators and functions will modify $_ in some way. Even if you don't use it, a function you call might, and it'll get changed. This leads to hard-to-debug code of the worst kind...

Reserve use of $_ to shell one-liners and map/grep blocks. The rest of the time assign your own variable.

[matrixmadhan]

Thanks

I should not have used the term 'directly' which means in any case with respect to $_; but that's not how it works.

Too impressive. :)

[TGI]

There is an implicit check for definedness when you do while (<FOO>) { }. See the perlop section on I/O Operators.

while (<FOO>) {
  # do stuff
}

# terminates if we get a line that evaluates to false.
while (my $l = <FOO>) {
  # do stuff
}

# only terminates at end of file (or error reading file)
while (defined(my $l = <FOO>)) {
  # do stuff
}

# Let's take advantage of $_ and be somewhat safe(r).
{ local $_;
  while (<FOO>) {
    # do stuff
  }
}