Re: [boost] Filesystem: remove_all should not follow symlinks

From: Beman Dawes (bdawes_at_[hidden])
Date: 2003-11-24 13:47:34

Peter Dimov wrote:

 >Beman Dawes wrote:
 >> At 05:53 PM 11/21/2003, Peter Dimov wrote:
 >
 >[about reference counted links]
 >
 >>> It doesn't make sense to talk about shallow or deep semantics here.
 >>
 >> That's usually correct, but there are a few query operations where the
 >> shallow/deep issue might matter, depending on the operating system.
 >> Size/date/time being examples.
 >
 >Do you have a specific OS in mind? Size is a particularly interesting
 >example. ;-)

No, except that size is problematic on many file systems, so it wouldn't 
surprise me at all if size for a reference counted link is sometimes 
troublesome.

That being said, Boost.Filesystem could support a create_link() function 
based on POSIX link() and Windows CreateHardLink(). While it would be 
portable between POSIX and Windows for files, for directories it would 
always fail on Windows, and might fail on POSIX (because the functionality 
is optional for POSIX directories.

I'm somewhat worried about providing a function with so many caveats. Also, 
no one has asked for it.

 >From C++ point of view it is reasonable to consider a symlink something
 >that when passed to std::fopen actually opens another file. Under this
 >definition .lnk does not appear to be a symlink.

Yes, that's what I've been thinking. So I don't think Boost.Filesystem 
should provide a function to create symlinks; it would be too non-portable.

OTOH, providing an is_symbolic_link() function which always returns false 
on Windows would allow users to write correct portable directory recursion 
programs similar to remove_all.

--Beman