Re: [boost] bpl: status of boost_python_friend_fixes?

From: David Abrahams (abrahams_at_[hidden])
Date: 2001-06-02 18:22:27

Phil,

I think you are probably right. Ralf Grosse-Kunstleve, Ullrich Koethe, and I
have been working out a way to address these problems. The friend approach
turns out to tread across the line of legality as far as the standard is
concerned. I would never have used it in the first place, if MSVC had
supported partial specialization/ordering. The solution which I expect will
be ready in the next week, is to revamp the library to use those features
everywhere except for MSVC, which will rely on the friend function hack.

Ralf has been doing most of the work on the code, though, so you might not
to rely on my assessment of the time before a fix is ready.

Regards,
Dave

----- Original Message -----
From: "Phil Austin" <paustin_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Saturday, June 02, 2001 12:00 PM
Subject: [boost] bpl: status of boost_python_friend_fixes?

>
> I  suspect that my troubles with BPL and KCC 4.0e and 4.0d may be related
> to the friend problem  discussed a on the list a few weeks ago
> (http://groups.yahoo.com/group/boost/message/12167).  Specifically,
> I'm getting  the following  with KCC  4.0e on Linux Mandrake 8.0:
>
> <ipc1 ~/install/boost_1_21_2/libs/python/src> KCC -G -o libpycpp.so *.o
> Recompiling extension_class.o
> "../../../boost/python/detail/extension_class.hpp", line 362: error:
>           identifier "py_extension_class_converters" is undefined
>       return
py_extension_class_converters(boost::python::type<T>()).to_python(x);
>              ^
>           detected during:
>             instantiation of "PyObject *boost::python::to_python(const T
&)
>                       [with T=boost::python::string]" at line 168 of
>                       "../../../boost/python/reference.hpp"
>             instantiation of "boost::python::ref
boost::python::make_ref(const
>                       T &) [with T=boost::python::string]"
>
> I'm not sure after reading  most  of  the messages  whether there's
> a workaround for this (or if I'm on the right track).  Any advice
> appreciated.
>
> Regards, Phil Austin
>
> To unsubscribe, send email to: <mailto:boost-unsubscribe_at_[hidden]>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>