Hi Jiri,
When did you check out from CVS ? This problem should have been fixed today
with CINT 5.14.70
Could you try again ?
Rene Brun
Jiri Masik wrote:
>
> Hi,
>
> I'm trying to compile the development version from the CVS as of
> today. The compiler can't compile a couple of rootcint
> generated files. G__Base2.cxx is one of them
>
> static int G__TObject_Warning_0_6(G__value *result7,G__CONST char
> *funcname,struct G__param *libp,int hash) {
> G__setnull(result7);
> ((const TObject*)(G__getstructoffset()))->
> Warning((const char*)G__int(libp->para[0]),
> (const char*)G__int(libp->para[1]),
> (va_list)G__int(libp->para[2]));
> return(1 || funcname || hash || result7 || libp) ;
> }
>
> g++ -pipe -Wall -fPIC -fsigned-char -Iinclude -DHAVE_CONFIG -I. -o
> base/src/G__Base2.o -c base/src/G__Base2.cxx
> base/src/G__Base2.cxx: In function `int G__TObject_Warning_0_6(G__value *, const char *, G__param *, int)':
> base/src/G__Base2.cxx:2840: ANSI C++ forbids casting to an array type
>
> which seems to be correct because va_list is defined as an array on my
> system. I am running Debian/linux on powerpc, gcc 2.95.3, libc 2.2.1.
>
> Please could you extend the recent changes in the vararg handling to
> account also for this definition of va_list. I don't know how to do it
> myself. Modifying (va_list) to (va_list*) seems to work with simple
> testing but I'm not sure if it is correct.
>
> Jiri
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:33 MET