Re: glibmm-2.6 and gtkmm-2.6 naming conventions

[Murray Cumming <murrayc murrayc com>]

On Wed, 2005-01-26 at 09:13 +0100, Cedric Gustin wrote:
> Murray Cumming wrote:
> 
> >>I will soon commit my win32-related changes to the CVS gtkmm (HEAD),
> >>including the project files for MSVC. But before doing that, I'd like to
> >>know what will be the naming conventions for glibmm and gtkmm 2.6 in the
> >>stable branch. Will the libraries be named libglibmm-2.6-*.so and
> >>libgtkmm-2.6-*.so,
> > 
> > No. And if we were going to do that then we would have already changed
> > the names to -2.5-*.so.
> 
> Ok, reverted to 2.4 for glibmm and gtkmm.
> 
> Now, I have an additional problem with the changes to gtkmmconfig.h.in 
> you made yesterday.

Does it break the build? Why?

>  I saw you now #undef HAVE_FLOCKFILE and 
> HAVE_FUNLOCKFILE. I don't believe these extra lines belong to 
> gtkmmconfig.h, as they are only used in the examples/demos. Wouldn't it 
> be better to modify config.h.in, and add some platform-specific #ifdefs, 
> something like

We already have HAVE_FLOCKFILE and HAVE_UNLOCKFILE in config.h. I put
them there because I thought that maybe the demo was #including some
other config.h, but it did not make any difference.

> ----- config.h.in -----
> 
> /* SUN Forte, AIX, and Tru64 have the problem with flockfile and
>     funlockfile - configure finds it but the compiler can not find it
>     while compiling demowindow.cc. undef HAVE_FLOCKFILE and
>     HAVE_FUNLOCKFILE for now, so that it builds on those platforms. */
> 
> #if !defined(_TRUE64) && !defined(_AIX)

I am not aware of any _TRU64 define, so this would not work. And it
would still be a workaround for a problem that should not be happening.

> #  undef HAVE_FLOCKFILE
> #  undef HAVE_FUNLOCKFILE
> #  undef GETC_UNLOCKED
> #endif

This is all very strange, as explained in the comments in demos/gtk-
demo/demowindow.cc.  For now, I'd like to use what works.

-- 
Murray Cumming
murrayc murrayc com
www.murrayc.com
www.openismus.com