You all know, I'm sure, about how platform-specific input and output translation can happen to the "\n" character (on some systems). That is, when reading from a real disk file, there is a sequence of one or more character codes that gets translated into a "\n" character when read (assuming you don't call binmode). The reverse happens on a write to a physical file.

But what do you do when you have to duplicate this functionality? Say you have a buffer full of bytes that came from somewhere other than a disk file (so you can't use binmode) and you want to split it into lines of text. How do you know what character sequence is used for line endings on a particular operating system?

One obvious way to do this is to use a hash keyed by the value of $^O. But this quite ugly, requires me to know what all the line endings are, and breaks when someone comes up with a new port of Perl to a different operating system.

Another way might be to write a "\n" to a file in text mode and read it back in in binary mode, but this requires the existence of a writable file (which is not guaranteed) and is slow.

Has anyone come up with a better way to do this than the two methods mentioned above?