This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: libstdc++ v3 wants more libio functions now


On Tue, Jul 25, 2000 at 05:03:45PM -0700, Mike Stump wrote:

> each machine can have a different value.  The default is int.

Right...

> freebsd, lynx, svr3, svr4, alpha, linux, openbsd, netbsd, c4x, fx80,
> bsd386, rtemself, iris5, pa, vms all seem to use 32 bit quantities.

Okay. MSVC, my primary 'platform' of comparison, defines it to be a 16
bit quantity, AFAICT. For those interested in how the problem came
about:

I had needed to kludge around the missing 'wstring' type in gcc-2.95.2
by defining it myself, and so I had chosen wchar_t for the template
parameterization. However, this was client code, and the server
was compiled on MSVC, and expected 16-bit character strings. I was
able to trivally change my typedef to use 'unsigned short', but I
actually won't have that luxury now that libstdc++-v3 appears to
define wstring itself.

I had looked into redefining WCHAR_TYPE, but couldn't ascertain what
exactly I should have done in order to make this work.

Are there any compiler features that will make the process of
user-definition of wchar_t possible?

> Also, I'd expect that you can run grep just about as well as I can...

Must you always be obnoxious?

Thanks to those who provided useful information like the 10646 specs,
et al.

m.

-- 
Programmer             "Ha ha." "Ha ha." "What are you laughing at?"
Loki Software                      "Just the horror of being alive."
http://lokigames.com/~briareos/                   - Tony Millionaire

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]