Hal Wine writes 30-JUL-1996: >At 11:58 AM on 7/29/96, dsmall@mitre.org wrote: > >>I set $/ to an unused character because otherwise searches would end with >>each line. Note that I didn't use an input-output loop, because if my $/ >>character is really unused, my program will read the entire file into $test >>in one try. > >To read a file completely, you should use > undef $/; >That works even if the file has the character you think is unused! Excellent advice. >PLEASE, use octal when you're looking for a bit pattern -- it's more >portable, and clearer: > >$LF = "\012"; >$CR = "\015"; >$DOSeol = "$CR$LF"; >$MacEol = "$CR"; >$UNIXeol = "$LF"; Not that this is not excellent advice (I think it is) but I am curious: is there some reason not to use hexadecimal or decimal sets? Would there be anything wrong with: $LF = "\x0a"; $CR = "\x0d"; or with: $LF = pack("c",10); $CR = pack("c",13); admittedly the decimal way introduces more typing but I think that the other number systems sometimes shy folks off. >Remember, "\n" does NOT mean ASCII CR!!!!! OK I remember. I am still curious though. Peter Prymmer pvhp@lns62.lns.cornell.edu