This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Patch: Enforce -Werror during bootstrap


 > From: Zack Weinberg <zack@codesourcery.com>
 > 
 > I think this rather reemphasizes my point - the set of warnings I see
 > on i686-linux is larger and only partially overlapping the set you're
 > getting on solaris.  This is a build from this morning, so before the
 > patch you posted to deal with a few warnings; some of the difference
 > is addressed by that patch.

Ok so it's slightly different.  I predicted this would be the case and
said in my last message that I would check it out.  Thanks for doing
it for me. :-)

BTW the stuff from the gt*.h header files are covered AFAICT by the
bypasses in my patch.  So what's really left?  I built a
cross-compiler and got the following warnings.  In two cases I didn't
get warnings where you claimed you did.  I'd appreciate it if you
could elaborate on those.


 > # Number of warnings per file:
 > #       4 insn-conditions.c
(4x) insn-conditions.c:???: warning: string length `???' is greater than ...
maybe fix-able, at worst bypass-able

 > #       3 gcc/tlink.c
tlink.c didn't warn for me in a cross-config

 > #       2 gcc/crtstuff.c
I don't think crtstuff.c or libgcc2.c get -Werror the way I wrote the patch.

 > #       1 gcc/ra-build.c
ra-build.c:401: warning: `long' switch expression not converted to `int' in ISO C
Fix-able

 > #       1 gcc/function.c
function.c:737: warning: signed and unsigned type in conditional expression
Fix-able

 > #       1 gcc/emit-rtl.c
emit-rtl.c:1852: warning: comparison between signed and unsigned
Fix-able

 > #       1 gcc/collect2.c
collect2.c didn't warn for me in a cross-config.

 > #       1 gcc/ada/tracebak.c
 > #       1 ada/b_gnatb.c
It's not clear to me yet whether Ada in fact would actually get
-Werror with my patch.  If so, I'd turn it off unless someone else
steps forward to clean it up.  I don't regularly build ada.

 > 
 > I continue to believe that -Werror is premature.
 > zw


Anyway, we're left with a handful of warnings, Some of which one could
fix and maybe one which might require a bypass (modulo the missing
ones I couldn't see.)

I hardly think this proves your point.  In fact, I think it proves
mine, i.e. that it's doable.  We're millimeters away from success.  If
you and perhaps others would help we can cross the goal line.

Would you be willing to handle the 5-6 nits on x86-linux-gnu to
accomplish this?

		--Kaveh
--
Kaveh R. Ghazi			ghazi@caip.rutgers.edu


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]