home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!charon.amdahl.com!amdahl!rtech!pacbell.com!sgiblab!swrinde!elroy.jpl.nasa.gov!nntp-server.caltech.edu!andy
- From: andy@cs.caltech.edu (Andy Fyfe)
- Newsgroups: comp.sys.3b1
- Subject: Re: Taylor uucp 1.4 gamma on 3B1 with TCP (long)
- Date: 28 Jan 1993 02:22:52 GMT
- Organization: California Institute of Technology
- Lines: 46
- Message-ID: <1k7g1sINNoll@gap.caltech.edu>
- References: <1993Jan20.045619.1333@d-and-d.com> <1993Jan21.123039.8504@ohare.Chicago.COM> <1993Jan23.050509.7515@d-and-d.com>
- NNTP-Posting-Host: kodiak.cs.caltech.edu
-
- In article <1993Jan23.050509.7515@d-and-d.com> dnichols@d-and-d.com (DoN. Nichols) writes:
- [Don raises valid questions about gcc, and whether or not the include
- files have been fixed by fixincludes. On some systems, like sun, this
- is essential, unless you use the -traditional option to gcc.]
-
- I ran fixincludes from gcc 1.42 on the include files on my system, which
- is 3.50. It found nothing that needed changed.
-
- I also ran fixincludes from gcc 2.3.3. It did find some changes. They
- were
- 1) All references to "vax" (#ifdef vax, etc) were changed to
- "__vax__". Fixincludes does not know to look for "mc68k",
- which is what the 3b1 defines. This is needed if you use
- the -pedantic option of gcc, since predefined names like
- "vax" and "mc68k" are not used.
- 2) All references to "#define NULL 0" had "#undef NULL" inserted
- before them (even when the #define was inside "#ifndef NULL").
- 3) <stdio.h> was modified to include <stdarg.h>. This would be
- needed if stdio attempted to give a prototype for v*printf.
- 4) The definition of size_t is <sys/types.h> had
- "#ifndef _GCC_SIZE_T", "#define _GCC_SIZE_T" and "#endif"
- inserted around it. This is probably to straighten out
- the interaction of <sys/types.h> with <stddef.h>.
- 5) Two comments in <sys/user.h> were botched. There are a
- pair of back to back comments, and fixincludes thought
- /*...*//*...*/ contained the C++ comment indicator //,
- so it created /*...*/*...*/*/, which isn't right.
- It doesn't appear that any of these changes is absolutely necessary.
-
- One change I'll make for the next version of gcc 2.x is to define
- SIZE_TYPE to be int, rather than letting it default to long unsigned
- int, so as to match the definition in <sys/types.h>. (Is it an int
- in other versions of the OS?)
-
- So far so good, but this only applies to 3.50.
-
- Could I enlist the aid of others who have different versions of the
- include files (either from a different version of the OS, or include
- files from another package) to check those as well? Then we can add
- a note to the gcc distribution about what might need to be changed,
- for future generations of 3b1 users.
-
- Andy Fyfe andy@cs.caltech.edu
-
- P.S. I've checked the uipc headers. They're fine, according to
- both versions of fixincludes.
-