Re: Problem with InstallShield (Was Re: [Bug 629] Changed - Problem with InstallShield: ole:CoTreatAsClass(stub), ole:CoGetClassObject)

On Wed, 8 May 2002, Dimitrie O. Paun wrote:

> On May 8, 2002 05:25 pm, Alexandre Julliard wrote:
> >
> > > There's another "excuse": proprietary hardware requiring custom core code
> > > under NDA (code that might go into, say, port.c). I've already mentioned
> > > my own hypothetical Xbox clone before, but TransGaming actually does wish
> > > to support some proprietary hardware in the future.
> >
> > As you say, that's just another excuse. If people can put the GPLed
> > Linux kernel on the Playstation proprietary hardware, there's no
> > reason you couldn't put the LGPLed Wine on it, or on any other piece
> > of hardware that you care about.
> 
> Indeed. I am surprised to see this sort of argument from Ove (whom I
> consider an exceptional programmer) which in fact is just Microsoft's
> argument that a modular Windows is not feasible... (which is ridiculous!)

Modularity and portability may go hand in hand, but they are not the same.
A lot of platform-specific stuff like the graphics and sound drivers can
be modularized out, but Wine uses many system features (you know all those
#ifdefs around Wine) that it would be counterproductive (or would kill
performance or something) to not allow inlining the NDA-ed implementation
of directly into core Wine code. I don't know how *much* code would suffer
really badly from getting such treatment, but there's always stuff that
you may want to put into a common .h file, or a file like port.c that's
used by all DLLs. There *will* be platforms where 100% system modularity
is a luxury you may not be able to afford. (Since Alexandre mentioned
Playstation: the PS2 has limited CPU resources, you probably want to
squeeze all you can out of it. Its power comes from the other chips in it.
Too many indirect calls on the main CPU in an effort to get around
licensing issues could be a FPS killer on this platform.)

I'm willing to agree that there will be few platforms and uses of Wine
that will suffer in practice, but I have never been willing to sacrifice
*any* use of Wine no matter how small. For all I care, the LGPL may make
good business sense, make us all richer than Bill Gates, and get us all
the code in the world, but it's *still* wrong to prohibit some uses of the
code, like non-modular adaption of the code for use on low-power embedded
systems, just because such use is not something we wanted to think about.
So ReWind will continue to exist.