Kenton Varda wrote:
> On Wed, May 20, 2009 at 12:58 PM, Monty Taylor <
> <>> wrote:
>     Works fine then, as I'm using it pretty consistently with Sun Studio -
>     guess it's just the pedanticness.
> Note that the whole "once" thing is new in 2.1.0 -- you've tested that
> one, right?

Ok. I reproduced (sorry - turns out I hadn't done 2.1.0 on my solaris
boxes yet - only on my linux boxes) and fixed.

The Sun autoconf check was assuming language and was checking for
__SUNPRO_C, but at some point an AC_LANG_PUSH(C++) had happened and
hadn't been popped. Solution: add an explicit PUSH and POP in the test.
:) The real key for the build failure is making sure -template=no%extdef
is in the CXXFLAGS

BTW, I'll put together a different patch that does the CXXFLAGS setting
stuff more cleanly. You are right - we should eventually be setting

diff -Nru protobuf-2.1.0/m4/acx_check_suncc.m4
--- protobuf-2.1.0/m4/acx_check_suncc.m4        2009-05-13
13:36:16.000000000 -0700
+++ protobuf-2.1.0-mt/m4/acx_check_suncc.m4     2009-05-21
19:57:41.917760080 -0700
@@ -12,7 +12,9 @@


-  AC_CHECK_DECL([__SUNPRO_C], [SUNCC="yes"], [SUNCC="no"])
+  AC_LANG_PUSH([C++])
+  AC_CHECK_DECL([__SUNPRO_CC], [SUNCC="yes"], [SUNCC="no"])

   AS_IF([test "$SUNCC" = "yes"],[
     isainfo_k=`isainfo -k`
@@ -30,4 +32,5 @@
     CFLAGS="-g -xO4 -xlibmil -xdepend -Xa -mt -xstrconst ${IS_64}
     CXXFLAGS="-g -xO4 -xlibmil -mt ${IS_64} ${MEMALIGN_FLAGS}
-xlang=c99 -compat=5 -library=stlport4 -template=no%extdef $CXXFLAGS"

You received this message because you are subscribed to the Google Groups 
"Protocol Buffers" group.
To post to this group, send email to
To unsubscribe from this group, send email to
For more options, visit this group at

Reply via email to