(I made the mistake of receiving the list as a digest)
Michal Hocko wrote:
> This is really weird. The code is CName("RGB") so there sis not
> CName(CName) in there. It looks like a compiler bug. Does the attached
> patch helps?
>
The patch:
image_dict.addProperty ("CS", CName((const char *)"RGB"));
at line 545 adding the cast as a constant character pointer yields the same
error before quitting.
What I know about c++ could be inscribed on the head of a pin leaving room
for the Gettysburg Address and angels doing a square dance. I'm a chip
designer by avocation.
A little experimentation:
The C shell allows commenting out continuation lines so I made a C shell
script with the compiler flags on separate lines (ending in '\') and
commented out one at a time with the same results.
Looking at boost:
%% find /usr/include -iname \*boost\*
/usr/include/c++/4.0.0/bits/boost_concept_check.h
/usr/include/c++/4.0.0/tr1/boost_shared_ptr.h
(gee that's not in macports /opt/local)
%% which g++
/usr/bin/g++
%% ll /usr/bin/g++
lrwxr-xr-x 1 root wheel 7 Jul 8 2009 /usr/bin/g++@ -> g++-4.0
Back to the issue of boost under macports:
%% find /opt/local/include -iname \*boost\*
/opt/local/include/boost
/opt/local/include/boost/fusion/adapted/boost_tuple
/opt/local/include/boost/fusion/adapted/boost_tuple/boost_tuple_iterator.hpp
/opt/local/include/boost/fusion/adapted/boost_tuple.hpp
/opt/local/include/boost/fusion/include/boost_tuple.hpp
/opt/local/include/gcc43/c++/bits/boost_concept_check.h
/opt/local/include/gcc43/c++/bits/boost_sp_shared_count.h
/opt/local/include/gcc43/c++/tr1/boost_sp_shared_count.h
/opt/local/include/gcc43/c++/tr1_impl/boost_shared_ptr.h
/opt/local/include/gcc43/c++/tr1_impl/boost_sp_counted_base.h
boost appears to have been built with gcc-4.3.4.
%% find /opt/local -iname g++\*
/opt/local/bin/g++-mp-4.3
/opt/local/share/man/man1/g++-mp-4.3.1.gz
/opt/local/var/macports/software/gcc43/4.3.4_0/opt/local/bin/g++-mp-4.3
/opt/local/var/macports/software/gcc43/4.3.4_0/opt/local/share/man/man1/g++-mp-4.3.1.gz
I switched to the /opt/local/bin/g++-mp-4.3 in the C shell and it compiled
with only notice of an unrecognized option '-posix' (which must be inherent).
The answer appears to be to switch to gcc-4.3.4. I'd suspect any
instability in running programs compiled using macport libraries might be
similarly attributable, as in previous versions of pdfedit.
Thanks Michal. Telling me it wasn't the code was enough to force a look at
the methodology.
Now you have to wonder if configure should check for naive users and
compiler/library matches?
It's been educational.
Thanks
------------------------------------------------------------------------------
_______________________________________________
Pdfedit-support mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pdfedit-support
