Re: gcc2 merge

[Jeffrey A Law <law at hurl dot cygnus dot com>]

  In message <199901281501.KAA18437@caip.rutgers.edu>you write:
  > 	Briefly looking over the changes, it looks like some bogus
  > stuff creeped in.
It's certainly possible.  All the error handling code was hit very hard by
these changes.  Combine that with all our work on the error handling routines
and we've got the potential for some stuff to sneak by.

The good news is we only have to go through this once for each major hunk of
code which outputs diagnostics.  That stuff doesn't tend to change a whole
lot over time.


  > 1.  I think I disagree with putting the PRINTF_PROTO_* stuff go into
  > system.h.  It was only used by cccp.c and cexp.y and it got nuked from
  > cccp.c.  I think we should deprecate this form and we should be using
  > the "PVPROTO ((...)) ATTRIBUTE_PRINTF" style instead.  This is a cleanup
  > issue, not a serous problem though. 
Fine.  Awaiting patch.


  > 2.  However, when PRINTF_PROTO_* was removed from cccp.c, the vararg
  > prototypes in that file no longer will get __attribute__ ((__printf__)). 
  > So they must have ATTRIBUTE_PRINTF_* added to their prototypes or no
  > format checks are done. 
OK.  I assume you'll fix that too ;-)



  > 3.  All the vararg prototypes in cccp.c incorrectly switched from
  > PRINTF_PROTO_* to VPROTO.  This is completely wrong since VPROTO is only
  > valid on function _definitions_ not prototype declarations.  One must
  > use PVPROTO in prototypes.  This mistake causes a bootstrap failure for
  > any stage1 compiler which doesn't default to ANSI.  I.e. OSF4 cc and
  > SunOS4 cc. 
  > 
  > 	I'll see about making a patch, but in the mean time, here is
  > the output from OSF4 cc.
Eagerly awaiting a patch :-)


jeff