Re: [boost] boost.filesystem - rationale for copy_file

From: Jody Hagins (jody-boost-011304_at_[hidden])
Date: 2005-01-27 16:40:06

On Thu, 27 Jan 2005 12:39:34 -0700
"Victor A. Wagner Jr." <vawjr_at_[hidden]> wrote:

> At Tuesday 2005-01-25 14:28, you wrote:
> 
> >What is the rationale for having copy_file fail if the target file
> >exists?
> >
> >This is not how any cp/copy in any shell I know of works, actually
> >quite surprising.
> 
> in windows (and every OS I've been on the writing team) the shell will
> at least _ask_ for permission to overwrite.

Hmmm.  I don't do windows, so I am no help there.  However, I am
somewhat familiar with the unix environment, and the "cp" command under
unix will copy over the existing file, without an error, warning, or
prompt.  If you want that behavior, you have to pass it a command line
argument (usually -i).  Some people may alias cp to get that behavior in
their interactive shells (e.g., alias cp="cp -i"), but it is definitely
not the default, and I agree that copy_file() call should not force the
file to be gone before doing to copy.