Subject: Re: newvers.sh padding problem
To: Curt Sampson <cjs@cynic.net>
From: Brian C. Grayson <bgrayson@marvin.ece.utexas.edu>
List: tech-toolchain
Date: 02/02/1999 20:27:36
On Tue, Feb 02, 1999 at 11:45:41AM -0800, Curt Sampson wrote:
> On Mon, 1 Feb 1999, Brian C. Grayson wrote:
> 
> >   Here's a different idea.  It's a gross hack, and relies upon
> > gcc.
> 
> Given that our last gross hack broke twice, prompting this whole
> discussion, I'm not very keen on putting another one in. :-)
> 
> I'd go for a char * alias, but aliasing version to be a function
> is not a good idea, I think.

  Hm.  According to the docs, only function aliases are
supported.  What if there was a big comment next to the
function alias saying we were doing a kludge?  :)

  Other ugly options include:  some sort of (m-i?) asm statement,
faking up a symbol table entry (a la cdefs.h?), changing the
symbol table entry for 'version' after the compile.  I can't
think of any tricky char[0] or struct things that would work.

  Or we could just give up those extra 120 bytes.  :)

  Brian