Abdelrazak Younes <[EMAIL PROTECTED]> writes:
> > | > I'd prefere is most of this work tried to leverage boost::filesystem
> > | > as much as possible.
> > |
> > | Yes, I think I've read somewhere that boost::filesystem will support
> > | the default encoding of the system in 1.34.
> >
> > Note that we are using 1.34 now.
>
> Yes and the support seems to be here indeed. See this for example
> extracted from the boost source code:
>
> inline bool is_directory( const path & ph )
> { return is_directory<path>( ph ); }
> inline bool is_directory( const wpath & ph )
> { return is_directory<wpath>( ph ); }
>
> The problem is that wpath is using wstring:
>
> typedef basic_path< std::wstring, wpath_traits > wpath;
>
> So I don't know how to make use of that for now. I guess using wstring
> is out of the question because we already have docstring.
No, all is as it should be. NTFS uses UTF-16 (thanks again, Joost), so that
can be stored reasonably well in std::wstring although there will be corner
cases where two wchar_ts are required to represent a single unicode character.
Think of std::wstring as the way to communicate with the native file system.
We're only going to do so on file input/output, at which points we're going to
have to convert between our internal use of docstring (UCS-4) to std::wstring
(UTF-16).
> I am getting a headache ;-/
Sure. No one said that this wasn't going to be a PITA.
Angus
