On Tue, Nov 19, 2002 at 07:07:46AM -0500, Klaus Klein wrote:
> Bang Jun-Young <junyoung@netbsd.org> writes:
> 
> > On Tue, Nov 19, 2002 at 11:32:22PM +1300, Dave Sainty wrote:
> > > Bang Jun-Young writes:
> > > 
> > > > 
> > > > Module Name:	basesrc
> > > > Committed By:	junyoung
> > > > Date:		Tue Nov 19 06:53:18 UTC 2002
> > > > 
> > > > Modified Files:
> > > > 	basesrc/lib/libc/string: strcasecmp.c
> > > > 
> > > > Log Message:
> > > > Include sys/types.h for u_char.
> > > 
> > > Wouldn't it be nicer just converting all the "u_char"'s in the file to
> > > "unsigned char"?
> > 
> > Is there any compelling reason to not include sys/types.h here and make
> > use of u_char? I did 'make build' after this change, and there was no
> > problem.
> 
> Your log message does not provide an accurate description of the
> change, however.  It does not mention that you included <sys/types.h>
> to supersede a local definition of u_char.  Also, "namespace.h" should
> precede _all_ other headers included (other than <sys/cdefs.h>),
> including <sys/types.h>.

I have done both (changing the log message accordingly and putting
sys/types.h after namespace.h).

> 
> I have to agree with Dave that having less of u_char would be much
> nicer.

What do you think about u_long and u_int?

Jun-Young

-- 
Bang Jun-Young <junyoung@netbsd.org>