[dfports] devel/boost - updated version.

[Henry Kleynhans <henry@xxxxxxxxxxxxxx>]

Hi all,

I've created an updated version of the dfports override for boost. 
Based on the information provided by Jeremy Messenger and Joerg 
Sonnenberger I think it would be better to keep the WCHAR support 
disabled until proper support has been ported to Dragonfly.

I've created a patch based on the code provided by Patrick Mauritz that 
changes the definition of BOOST_PYTHON_OFFSET if we are compiling with a 
GCC compiler with a version greater than 2 on DragonFly.  Are we still 
using an older version of the GCC 3 compiler on DragonFly that could 
influence this?  Here is that patch:

--- boost/python/detail/config.hpp~     2004-09-26 22:20:13.000000000 +0200
+++ boost/python/detail/config.hpp      2004-09-26 22:19:55.000000000 +0200
@@ -104,6 +104,9 @@
 // Replace broken Tru64/cxx offsetof macro
 # define BOOST_PYTHON_OFFSETOF(s_name, s_member) \
         ((size_t)__INTADDR__(&(((s_name *)0)->s_member)))
+#elif defined (__DragonFly__) && defined(__GNUC__) && __GNUC__ > 2
+#define BOOST_PYTHON_OFFSETOF(s_name, s_member) \
+       (__offsetof__ (reinterpret_cast <size_t> (&reinterpret_cast 
<const volatile cha
r &> (static_cast<s_name *> (0)->s_member))))
 #else
 # define BOOST_PYTHON_OFFSETOF offsetof
 #endif


I've tested the dfport override with the GCC34 on DragonFly CURRENT. 
Everything now compiles including the python library.  I haven't done 
exhaustive testing to make sure that everything works.  I've done some 
testing on the python library and that seems fine.

It would be great if someone can have a look at this.  Maybe also try 
and compile it with gcc2 and see if that works.

Kind regards,

Henry

Attachment: boost-dfports-override.tar.bz2
Description: application/bzip