Angus Leeming <[EMAIL PROTECTED]> writes:
| Lars Gullik Bjønnes wrote:
| > | > | We use char_type, not size_t for storing UCS4 characters.
| > | > | This may make a difference when char_type is signed.
| > | > When is char_type signed?
| > | | http://www.network-theory.co.uk/docs/gccintro/gccintro_71.html
| > This does not apply (afaics)
| > We never use 'char' for char_type.
|
| My bad. I should have read the patch.
|
| #if defined(HAVE_WCHAR_T) && SIZEOF_WCHAR_T == 4
| // Prefer this if possible because GNU libstdc++ has usable
| // std::ctype<wchar_t> locale facets but not
| // std::ctype<boost::uint32_t>. gcc older than 3.4 is also missing
| // usable std::char_traits<boost::uint32_t>.
| typedef wchar_t char_type;
| #else
| typedef boost::uint32_t char_type;
| #endif
|
| A quick google search for 'signed wchar_t' turns up this:
| http://mail.nl.linux.org/linux-utf8/2000-08/msg00107.html
I am not sure that we should even try to support signed wchar. Perhaps
a test for that is in order.
--
Lgb
