Re: [sigc] looking for mailing list archive

[Martin Schulze <martin-ml hippogriff de>]

Hi Rudi!

Sorry for the late answer - I've been on a course for the last two  
weeks and didn't spot you email in the webmail interface.

Am 28.09.2004 14:01:16 schrieb(en) Rudi Wurm:
> i'm coming from the libsigc++ website, have the problem compiling
>
>   libsigc++-2.0.1 to libsigc++-2.0.5
>
> with
>
>   gcc 3.2.2
>
> In file included from test_functor_trait.cc:7:
> ../sigc++/adaptors/bind.h:1049: `deduce_result_type_internal' is not  
> a
> template
> ../sigc++/adaptors/bind.h:1049: `type' is not a class or namespace
> ../sigc++/adaptors/bind.h:1049: ANSI C++ forbids declaration `type'
> with no type
>
> tried -ansi and -traditional-cpp flags
>
> looking for solution ...

libsigc++2 definately compiles with gcc 3.3.x. It should also compile  
with gcc 3.2, however since I cannot test every platform myself I have  
to ask for your patience in this case.

First step: please open a bug report at bugzilla.gnome.org for this.  
Attach the make output for libsigc++-2.0.5 or current cvs (module  
libsigc++2).
Second step: This is obviously a compiler bug so we need a work-around  
and a test case for configure. The error occurs in code generated from  
sigc++/adaptors/macros/bind.h.m4, line 137 (in cvs) so this is the one  
to play with. A random idea would be to remove the "typename" keyword.

Regards,

 Martin