Re: Results for egcs-2.92.19 19981108 (gcc2 ss-980609 experimental) testsuite on i586-pc-linux-gnu

["Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>]

 > From: Nix <nix-egcs@esperi.demon.co.uk>
 > 
 > It looks like Kaveh's change to splay-tree is at fault:
 > 
 > Sun Nov  8 17:42:25 1998  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
 > 
 >         * ansidecl.h: Wrap problematic macros with !IN_GCC.
 > 
 >         * demangle.h: Never define PARAMS().
 >         * splay-tree.h: Likewise.
 > 
 > The problem here being that splay-tree, of course, still *uses*
 > PARAMS. PARAMS is, of course, defined in ansidecl.h; unfortunately,
 > nothing in ansidecl.h is being included, because IN_GCC is defined
 > when we're in GCC (surprisingly ;) ).
 > 
 > Why is ansidecl.h being excluded in this case? Should it be?



	There was a small window of time while I was doing a checkin
where you wouldn't have had all the necessary changes.

	Try doing another checkout and see if that helps.  Make sure
you see the following in egcs/gcc/ChangeLog:


 > Sun Nov  8 17:50:30 1998  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
 > 
 >         * gansidecl.h: Prepend a "G" to the macro wrapping this file
 >         (to distinguish it from the macro wrapping ansidecl.h.)
 >         Include libiberty's ansidecl.h.  Remove all redundant definitions.
 >         Define the PROTO() style macros in terms of the PARAMS() ones.


	As the entry explains, gcc's gansidecl.h now includes
libiberty's ansidecl.h.  This was done to force conflict resolution
between the two which allows closer merging of gcc with libiberty.

The specific long term goal is to include libiberty.h in gcc source,
and remove duplicate stuff from gcc which libiberty.a provides.
That's in the works, stay tuned...

		--Kaveh
--
Kaveh R. Ghazi			Engagement Manager / Project Services
ghazi@caip.rutgers.edu		Icon CMT Corp.