On Mon, 11 Feb 2002, Jerrad Pierce wrote: > Alan Jaffray: > > > >Would it be better language design to actually define sensible and > >portable behavior for functions that do OS-ish things, rather than > >pass them along the underlying OS and wash our hands of them? Well, > >yeah. But it's a legacy we're stuck with, at least for Perl5... > > I'd have to disagree. If one knows what rename does, > which is simply to play with the filesystem and not touch the data, > then it is what you want if you know that that is what it is doing.
OK. I realize what I wrote sounds like "CORE::rename should behave the way File::Copy does now", and that's not what I meant. IMHO neither File::Copy nor CORE::rename should be in core. Someone who knows what rename(2) does and wants that behavior can use a module to get it. Having that behavior in core with the tempting title "rename" is like putting a big red START button on a machine with a note in the manual saying "button may or may not cause machine to activate; if you're thinking about pressing that button, you probably want to flip a switch on the underside instead." :) > As for Perl6 the plan is for most things to not live in the core anyways. And when we have Perl6, what a wonderful world it will be... Alan
