Strategy (was Re: Improving copyright)

[Jeroen Dekkers]

At Fri, 14 May 2004 12:18:55 +0100,
Niall Douglas wrote:
> On 13 May 2004 at 22:55, João Miguel Neves wrote:
> 
> > > What mistakes? Windows NT was designed by one of the most famous &
> > > respected operating system designers the world has ever known. It
> > > has a superb kernel. While it has a few structural mistakes, it has
> > > far fewer than say the 2.4 series Linux kernels [1] and remember it
> > > was designed long before Linux.
> > >
> > You completely misunderstood me. Any program with a thousand of lines
> > has hundreds of little choices that were made by it's programmers. A
> > clone like ReactOS has to get all those right. It's that task that I
> > think it's mad.
> 
> Linux has had this problem twenty times over in trying to behave like
> all the different Unices out there *at* *the* *same* *time*.
> Following this logic, Linus should have dumped the idea of Linux and
> gone with forking Plan 9 instead - which while very appealing on
> technical grounds, it ignores how much worth legacy compatibility is
> worth to people.

GNU/Linux almost complies to an international standard called
POSIX. Why would you want to throw away that standard and implement
some non-standard, proprietary API which changes every few years?
Don't you like open standards?
 
> > I don't think that running windows binaries is going to be an issue in
> > less than one decade, probably 3 to 5 years.
> 
> Precisely why Microsoft are going down the managed code route.
> They're going to try and get all new development over to that new
> proprietary API with proprietary tools & programming languages. By
> the time Linux is fully up to par, all the latest apps will be
> incompatible again.

Linux is up to par with what? Do you mean wine instead of Linux?
 
> > > Far better clone that system [2] and beat the devil at his own game.
> > >
> > No, running in second place by following the person that goes first is
> > not beating "the devil at his own game". It's making sure you're
> > always behind the first guy. Our resources are better spent on
> > platforms that have shown themselves stable enough over the years that
> > don't force us to reimplement our programs every 3 years.
> 
> Are we talking about the same system here? I can run my Win95 and
> often my DOS binaries unmodified on the latest Windows. A Linux
> binary from 1996 stands *zero* *chance* of running unmodified on the
> latest Linux.

I know from other people that most games written for win 9x work
poorly or not at all on XP. Even microsoft can't make XP fully
compatible with 9x, and we should be able to do so without specs,
source code and knowledge of the internals?  I don't think so.
 
> When you say "It's making sure you're always behind the first guy" I
> think you're ignoring how non-techies value computers. Non-techies
> want something they can buy some peripheral or application for with
> reasonable assurance it'll work - Linux doesn't and never will do
> this - it has more of the market now than MacOS X yet its peripheral
> support is far less. That's because Linux is good for techies (who
> are able to maintain their own kernel modules) and servers (which
> have a very limited set of hardware configurations).

Linux will never do that? And how do you know? Can you look into the
future?

> Now if you had a free clone system capable of running 100% Win32
> binaries and drivers natively you remove the obstacles preventing non-
> techies adopting it instead of MS Windows. If even 15% of the market
> use clone Windows, suddenly application developers must ensure
> compatibility - which means NOT using any new proprietary features
> tacked on like managed code (actually, if clone Windows were really
> good enough, you could install Microsoft's own .NET runtime :) ).
> 
> So you see what I'm driving at? Linux encourages MS to keep pushing
> people to use new proprietary systems. A binary compatible Windows
> clone makes such tactics redundant.
>
> I'm not saying drop Linux tomorrow - I AM saying the FSF should adopt
> ReactOS and throw resources at it if it's really serious about
> creating change. Of course, this will unleash every weapon MS has
> got, but better do it now than in ten years when they'll have
> finished patenting everything Linux is made of.

Which resources? FSF doesn't have any resources for this at all. And I
think the FSF had created enough serious change and is still doing so
- or do you want to say the FSF didn't create a change the last 20
years? Do you want to say that the increasing number of GNU/Linux
users isn't a change?

And about the patenting stuff, I don't think there will be software
patents in Europe.

Jeroen Dekkers