From: Edward Diener (eddielee_at_[hidden])
Date: 2004-02-15 22:48:53
Jeff Garland wrote:
> On Sun, 15 Feb 2004 15:37:14 -0500, Edward Diener wrote
>> Paul Miller wrote:
>>> I just started using boost recently and it seems excellent so far.
>>> One of the first classes I started using is filesystem::path, with
>>> the intention to use it portably on Windows, OS X, and Linux.
>>> On Windows and OS X, you can work with paths in Unicode. Presumably
>>> Linux still works with multi-byte characters.
>>> Is there progres toward a wchar_t-aware path?
>> I don't know if Mr. Dawes will include such an addition in a future
>> version of filesystem, but I do know that urging that C++ standard
>> add wide character filenames and paths where appropriate, on
>> comp.std.c++, has met with a wall of resistance in the past. There
> Of course, given that boost is open source, there is always another
> path. The users that really want wchar_t support can dig in and work
> on coming up with a design / implementation. I'm sure Beman and
> others would welcome such contributions and are willing to help
> discuss the design options.
> Having just finished (checked in today) adding wide string/stream
> support into date_time, I can tell you it is a non-trivial issue.
> The fundamental problem traces back to those legacy compiler/library
> combinations. My goal wasn't to support wide string/streams on the
> legacy platforms, only to not break the current level library support
> for those compilers. That was tricky without hacking up the code
> base too badly. So I would council patience on the part of folks
> that aren't willing to spend the time to work out a concrete proposal
> for how to approach the problem.
I think there are two separate issues which shouldn't be mixed, and I don't
know if you are inadvertently mixing them by your remark. One is support for
wide character file names, and the other is support for wide character file
streams. The C++ standard library does support wide character file streams,
it just doesn't support wide character file names.
I wasn't in any way criticizing filesystem for not having support for wide
character file names, or even supposing that it should support wide
character file names. In fact I don't think it should unless the C++
standard library supported wide character file names in its basic_fstream
template classes, which it doesn't.
As far as a proposal for the C++ standard library supporting wide character
file names, I did not make one for a number of reasons, one of which is that
the feeling against it, from comp.std.c++, of leading C++ standard committee
members was so strong that I did not feel it to be worthwhile even though I
disagreed with them and argued my disagreement there.
Finally I think that adding wide character file name support to the C++
standard library is a much different issue from adding it to filesystem,
with adding it to filesystem being a much more complicated endeavor than
adding it to the C++ standard library classes. But that may be a discussion
which really doesn't belong here, since it doesn't really deal with
filesystem but with my opinion about how to add wide character file name
support to the C++ standard library.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk