Re: [Boost-users] wide char i/o problem.

From: Robert Ramey (ramey_at_[hidden])
Date: 2007-08-15 18:10:21

David Abrahams wrote:
> on Tue Aug 14 2007, "Robert Ramey" <ramey-AT-rrsd.com> wrote:
>
>> You'd have to talk with the mingw developers.
>>
>> To use gcc on windows, I use cgwin which comes with gcc.  A very
>> good pacakge - but alas, it doesn't support wide characters either.
>
> Robert,
>
> There are some fairly obvious questions you're not answering.  For
> example, does this error make Boost.Serialization unusable with gcc on
> windows, and if not, how does one make it work?

It makes makes archives based on wide character streams unusable
with the versions I've seen for gcc that work on window.  Other
archives work just fine.  If one needs to use an archive which
is based on wide character i/o (ie wstreambuf) on windows,
he will have to procure such an implementation.  I don't have
one and I don't know where to get one.  Sorry

Perhaps there is a little bit of confusion in the following.  All
archives included in the libray support i/o wide characters
and wstring.  When loading/storing wide characters on
normal character archives, translation between unicode
and multibyte code base on the current locale is made on the fly.
Thus any archive will hold any datatypes.

If one wants to use some other locale for this purpose
it can be tied to the stream in the normal way before
he archive is created.

unicode<->multi-byte translation is slightly fragile due
to the requirement of picking the correct locale.  But
if you don't have wide character i/o available, its your
only choice.

Robert Ramey