RE: [boost] Filesystem attributes

From: Jan Langer (jan_at_[hidden])
Date: 2002-03-06 08:17:08

On Tue, 5 Mar 2002, Stewart, Robert wrote:
>I would expect "0".  I would expect that f1 and f2 rely on the same
>underlying data structures (your file_rep).  Thus, "set(user_read, *f2,
>false);" would flush all pending changes.  Likewise, any iterator that
>refers to the same file should refer to the same file_rep.

how should this be guaranteed? eg. symbolic links, the same filessystem
mounted in two different ways (smb and nfs, although i never tried
this), a local file and a net ressource representing my local file.

>Now, having said that, because of the complexities and even performance
>issues that might result from those expectations, one could, just as easily,
>decide that referencing the same file's attributes via iterators and
>multiple dir_entry objects at the same time (I'm not talking about threads,
>though that's a whole separate issue), is undefined.

i agree to this.

-- 
jan langer ... jan_at_[hidden]
"pi ist genau drei"