> On Saturday 05 May 2001 19:33, Andrew C. Greenberg wrote:
> 
> > Recent arguments made by Microsoft executives criticizing open source
> > development criticize such systems point to forking software as its
> > achilles heel.
> 
> As someone who's tried to port things between different versions of Windows 
> (even Win32) over the years, I'd have to say that Microsoft is no stranger to 
> the forking problem.

Amen to that.  As someone who has to test software regularly between all 
variants of Windows (even...ick...Win 3.1) I can wholeheartedly attest to the 
non-uniformity between Windows flavours.

Microsoft ranting about forking being "bad" is quite humourous...but I doubt 
that the VP in question has actually done much work trying to -code- between 
the different incarnations of Windows either.


> 
> For some reason, for instance, their Win9x and WinNT groups didn't bother 
> coordinating where things go in the registry.
> 

And don't forget WinME with it's totally "new" way of dealing with environment 
variables (no, they couldn't leave autoexec.bat well enough alone, nor could 
they impliment something like NT's environment variable system...)

> Or you could look at a variety of Win32 API calls that work just enough 
> differently to be annoying.
> 
> For entertainment, look at MFC source code sometime. There is (or was the 
> last time I looked) considerable conditional compilation depending on OS 
> version.
> 

You find MFC entertaining? :)

> It's not at all uncommon to see software vendors supplying a Win9x version as 
> well as a WinNT version of the same _application_ package.
> 
> -- 
> Ned Konz
> currently: Stanwood, WA
> email:     ned at bike-nomad.com
> homepage:  http://bike-nomad.com