>>while (<>) {
>> tr/\015\012//d;
>> ...
>>}
>>
>>And that would work with CR or CRLF (or under unix with LF or CRLF).
>
> you should instead
>
>while(<>){
> chomp;
>}
>
> Takes care of any combination of CRLF, CR and LF.
No it doesn't, at least not according to the docs at
http://www.perldoc.com/perl5.6/pod/func/chomp.html
chomp will only remove the $/ character which is typically \015 for
MacPerl and \012 for unix perl. The tr code, while definitely
slower, will work with CRLF as well in either unix or MacPerl.
> Unfortunately $/ trick doesn't work on all combination of CRLF, CR
>and LF. One possible solution would be as follows;
>
>undef $/; # enable "slurp" mode
>my $content = <FH>; # whole file now here
>chomp $content; # trailing CRLF would be gone.
>for my $line (split /\015\012|\015|\012/, $content){
> # now process $line one by one.
>}
Yes, something like that would work, and as you say, memory is
usually not that relevant.
> I think the best approach toward this is to write a module that
>does that. IO::Handle::AnyNewline?
Yeah, something like that was what I was thinking. Although again,
it is a lot more mess than just <>. It seems to me it's worth
another special case for $/, although that is running out of special
case possibilities! It's never been a problem before because I never
had to deal with Mac & Unix files, it was always one or the other,
possibly plus DOS/Network CRLF files. Oh well. It's a shame $/ is
not a regex, but I guess that runtime cost of that would be a bit
much.
>P.S. Thank you for great products like anarchy and Interarchy.
>Though I am not a user myself my customers have benefit a lot.
Glad to hear it!
Thanks for your help,
Peter.
--
<http://www.interarchy.com/> <ftp://ftp.interarchy.com/interarchy.hqx>
