solaris bootstrap failure using cc for stage1, stab.h not found...

[Kaveh R. Ghazi]

 > From: Jeffrey A Law <law@hurl.cygnus.com>
 > 
 >   In message < 199812071939.OAA21225@caip.rutgers.edu >you write:
 >   > 	I suspect the following patch will do the trick.  It uses the
 >   > same strategy to protect gstab.h/stab.h as found in final.c and
 >   > xcoffout.c.  Using this patch on Solaris, I am able to compile dbxout.c
 >   > in stage2 (having used cc for stage1.)
 >   > 
 >   > 	Okay to install?
 >   > 
 >   > 		--Kaveh
 >   > 
 >   > 
 >   > Mon Dec  7 13:59:11 1998  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
 >   > 
 >   > 	* dbxout.c: If USG is defined use gstab.h, even if HAVE_STAB_H is set.
 > It's fine.  Particularly since it makes dbxout.c match how other parts of
 > the compiler work.

	Installed.


 > I'd still like to think about just using our gstab.h long term and doing away
 > with the conditionals to select different stab.h files.  Consider cross
 > compilers for example.  We don't want to use the system stab.h in such cases
 > (maybe we already have code to avoid that...  I haven't looked).
 > jeff

	I think simply ripping out <stab.h> from gcc would work. 

Consider, before Andreas' patch on Nov 9th, gcc was using the incorrect
macro HAVE_STABS_H (i.e. incorrectly pluralized STAB) to determine
whether the OS' stab.h should be used.  This had the effect of making
*every* platform use gstab.h since HAVE_STABS_H (with the plural STAB)
would never be defined.  I suspect this typo had been present ever since
gcc was autoconfiscated (between 2.7 and 2.8 ?)

	However, I'll leave it to someone better versed in dbx debugging
and cross compilers to take the step of removing the OS/stab.h. 

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