of NULL was incorrect for c++ compilers that aren't
gcc (or pretend to be gcc like clang).
ok miod@

Kernel and bootblocks still use the old 0L value until all the NULL abuses
in the code are fixed.

This is required as gcc grows ever more pedantic with old age
and cavorts with standards bodies that like to create paradoxes.
ok kettenis@ miod@ millert@ espie@

Add new sys/_types.h header
Include machine/_types.h or sys/_types.h where applicable

* introduce the mbstate_t typedef.
impacts gnu libiconv, which has already been taken care of.
* Prepare for mb stuff to really exist, replace macro MB_CUR_MAX with
an external variable __mb_cur_max (impacts libX11 and various ports).
* use mbstate in all the mb <-> wchar functions with state.
* add a stub iswctype function allowing some ports to compile.
bash and gdiff are missing wcscoll, and need to be told there's no i18n
until this is fixed.
Discussed and matthieu, otto, millert, kettenis, deraadt.
Major libc bump

okay deraadt@, cloder@, pvalchev@
thanks to everyone who commented on that.

rescinded 22 July 1999.  Proofed by myself and Theo.

OK mickey@ and discussed with deraadt@

Redefining it as a typedef is an error, as it prevents overloading
on e.g., int/wchar_t separately.

__null is a magic constant of integral type that converts to a null pointer
as should be, but warns for ambiguity when used to resolved an overload
between an integral type and a pointer type.

entire trees for testing anyway, I might as well do this intrusive touching
of include files now.  Added openBSD tags.