🚀 go-pugleaf

Roland McGrath writes:

     There is a potential problem with the `size_t' type and versions of
  GCC prior to release 2.4.  ANSI C requires that `size_t' always be an
  unsigned type.  For compatibility with existing systems' header files,
  GCC defines `size_t' in `stddef.h' to be whatever type the system's
  `sys/types.h' defines it to be.

That is all correct.

  Most Unix systems that define `size_t'
  in `sys/types.h', define it to be a signed type.

I don't believe that this is completely accurate.  Looking at the files in
the current development (pre-release) version of GCC 2.3, I see the following
definitions:

a29k.h:#define SIZE_TYPE "unsigned int"
convex.h:#define SIZE_TYPE "unsigned int"
dec-osf1.h:#define SIZE_TYPE    "long unsigned int"
decrose.h:#define SIZE_TYPE     "long unsigned int"
decstatn.h:#define SIZE_TYPE "unsigned int"
encrose.h:#define SIZE_TYPE     "long unsigned int"
hp320.h:#define SIZE_TYPE "unsigned int"
i386iscgas.h:#define SIZE_TYPE "unsigned int"
i386linux.h:#define SIZE_TYPE "unsigned int"
i386mach.h:#define SIZE_TYPE "long int"
i386sco.h:#define SIZE_TYPE "unsigned int"
i386sun.h:#define SIZE_TYPE "int"
iris.h:#define SIZE_TYPE        "unsigned int"
m88k.h:#define SIZE_TYPE "unsigned int"
mot3300.h:#define SIZE_TYPE "unsigned int"
news.h:#define SIZE_TYPE "long int"
news3.h:#define SIZE_TYPE "long int"
news3gas.h:#define SIZE_TYPE "long int"
pa-ghpux.h:#define SIZE_TYPE "unsigned int"
pa-hpux.h:#define SIZE_TYPE "unsigned int"
pa.h:#define SIZE_TYPE "unsigned int"
plexus.h:#define SIZE_TYPE int
romp.h:#define SIZE_TYPE "long int"
seq386.h:#define SIZE_TYPE "int"
sequent.h:#define SIZE_TYPE "int"
sparc.h:#define SIZE_TYPE "int"
sun2o4.h:#define SIZE_TYPE "int"
sun3.h:#define SIZE_TYPE "int"
svr3.h:#define SIZE_TYPE "unsigned int"
svr4.h:#define SIZE_TYPE "unsigned int"
ultrix.h:#define SIZE_TYPE "unsigned int"
vax-vms.h:#define SIZE_TYPE "unsigned int"
we32k.h:#define SIZE_TYPE "unsigned int"

Now I admit that this is not a "scientific" survey of all UNIX systems,
but it does make it appear that the majority of GCC target systems *are*
already conforming to the ANSI C standard with respect to the signedness
of size_t.  Only a few archaic systems do otherwise.


// Ron ("Loose Cannon") Guilmette
// uucp: ...uunet!lupine!segfault!rfg
// New new motto:  Quality control is a state of mind.
//   misc.forsale.computers ad, circa 2007:
//       Used Cray wrist watch for sale; 25 bucks or best offer.