On Wed, Sep 20, 2006, Doug Summers wrote:
> I know, it's really old, but I'm now stuck with about 20 of them to
> support and would really like to leverage OpenPKG. When trying to build
> OpenPKG-stable I get this during the building of RPM:
>
> source='depends.c' object='depends.lo' libtool=yes \
> depfile='.deps/depends.Plo' tmpdepfile='.deps/depends.TPlo' \
> depmode=aix /bin/sh ../depcomp \
> /bin/sh ../libtool --mode=compile /bin/cc -DHAVE_CONFIG_H -I.
> -I. -I.. -I. -I.. -I../build -I../rpmdb -I../rpmio -I../popt -DOPENPKG
> -DOPENPKG_AIX -I/tmp/openpkg-2.20060824/zlib-1.2.3
> -I/tmp/openpkg-2.20060824/bzip2-1.0.3
> -I/tmp/openpkg-2.20060824/beecrypt-4.1.2 -DOPENPKG -DOPENPKG_AIX
> -I/tmp/openpkg-2.20060824/zlib-1.2.3
> -I/tmp/openpkg-2.20060824/bzip2-1.0.3
> -I/tmp/openpkg-2.20060824/beecrypt-4.1.2 -c -o depends.lo `test -f
> 'depends.c' || echo './'`depends.c
> /bin/cc -DHAVE_CONFIG_H -I. -I. -I.. -I. -I.. -I../build -I../rpmdb
> -I../rpmio -I../popt -DOPENPKG -DOPENPKG_AIX
> -I/tmp/openpkg-2.20060824/zlib-1.2.3
> -I/tmp/openpkg-2.20060824/bzip2-1.0.3
> -I/tmp/openpkg-2.20060824/beecrypt-4.1.2 -DOPENPKG -DOPENPKG_AIX
> -I/tmp/openpkg-2.20060824/zlib-1.2.3
> -I/tmp/openpkg-2.20060824/bzip2-1.0.3
> -I/tmp/openpkg-2.20060824/beecrypt-4.1.2 -c -M depends.c -o depends.o
> "depends.c", line 48.12: 1506-343 (S) Redeclaration of rpmNAME differs
> from previous declaration on line 39 of "./rpmlib.h".
> "depends.c", line 48.12: 1506-382 (I) The type "unsigned char*" of
> identifier rpmNAME differs from previous type "const unsigned char*".
> "depends.c", line 51.12: 1506-343 (S) Redeclaration of rpmEVR differs
> [...]
> The compiler version is VisualAge 5.0 (can't get anything newer). I
> might be able to get a 3.4.x version of gcc built to try that.
>
> Any idea where I should look?
The above errors like strange. Just for an example:
lib/depends.c:48: const char *rpmNAME = PACKAGE;
lib/rpmlib.h:51: extern const char * rpmNAME;
Both look just fine and fully in sync to me. But what scares me is the
message about 'identifier rpmNAME differs from previous type "const
unsigned char*"'. From where does the "unsigned" come here? Anyway, the
compiler looks rather picky to me and the older RPM 4.2.1 sources are
not the optimal ones when it has to be passed by picky compilers as
experience showed. So, yes, I think you should give the GCC 3.4 a try,
please.
Ralf S. Engelschall
[EMAIL PROTECTED]
www.engelschall.com
______________________________________________________________________
The OpenPKG Project www.openpkg.org
User Communication List [email protected]
