Re: compile wrapper script with C++ file

[Ralf Wildenhues]

Hi David, and sorry for the long delay,

* David Byron wrote on Fri, Nov 13, 2009 at 08:18:35PM CET:
> On Friday September 4th, 2009, Eric Blake wrote: 
> > According to David Byron on 8/14/2009 1:57 PM:
> > > I'm not sure whether this is something I'm doing wrong
> > > or something wrong with automake, autoconf or libtool.
> > > At the moment I'm leaning to autoconf + automake.
> >
> > compile is maintained by automake, so I've redirected your
> > patch there.  FWIW, it looks okay to me, but Ralf will
> > have to chime in.

As David already correctly noted, merely changing the compile script
is not sufficient; it would also need some AM_PROG_CXX_C_O macro.
I have some unfinished patches from long ago for that (C++ and Fortran)
for both Automake and Autoconf (which also needs likewise additions).

However, I never finished them, for they'd only really help some SCO C++
compiler, and a few Fortran compilers, but it still wouldn't help with
MSVC: first, it *has* an /Fo$object switch, then there are still lots of
other things that won't work, like /Fe$executable.  For most of these
issues, you can use a wrapper script like cccl, or cegcc, or a couple of
others I forgot now.  But then, you don't need above changes, and the
least you want on w32 with slow fork emulations is another wrapper
script (the `compile' one) sitting in between make, the compiler,
possibly libtool, and a wrapper script or compiler.

I'm not sure about the best way to proceed here, or what level of
emulation to add to `compile', but going just one step seems less
attractive than the current status quo or going even further.

Hope that helps.

Cheers,
Ralf