Re: Design for a file system library

From: bill_kempf (williamkempf_at_[hidden])
Date: 2002-03-08 12:59:51

--- In boost_at_y..., "Stewart, Robert" <stewart_at_s...> wrote:
> From: Rainer Deyke [mailto:root_at_r...]
> > 
> > ----- Original Message -----
> > From: "Jan Langer" <jan_at_l...>
> > To: <boost_at_y...>
> > Sent: Thursday, March 07, 2002 2:28 PM
> > 
> > > actually i want no root_directory () function at all, but a
> > > hierarchy_iterator returning '/' for posix as a single element 
and
> > all
> > > drive letters beginning with 'a:' (if it exists) for win.
> > 
> > That would mean that recursive iteration will find printers under 
UNIX
> > (/dev/*) but not under Windows (where they are not under any drive
> > letter).  Is this intentional?
> 
> Since printers and other devices appear in the filesystem as files, 
yes,
> they should be accessible via such iteration.  A *nix developer 
should be
> able to get to everything in their filesystem.
> 
> The question becomes whether a Windows developer should have access 
to all
> of the things that can appear in "Desktop" or whether this library 
should
> restrict itself to files and directories.  I don't think non-file,
> non-directory objects should be accessible via a directory_iterator.

I'm not sure I agree, but in any event on DOS/Windows systems it's 
not quite so easy.  COM and LPT ports can be opened by file I/O 
operations, for instance.

Bill Kempf