Re: If Linux is to succeed

Ingo Molnar (mingo@pc5829.hil.siemens.at)
Sat, 25 Jan 1997 19:44:14 +0100 (MET)


On Fri, 24 Jan 1997, Alan Cox wrote:

> > Quote from my boss: "What the hell is taking [the WINE project] so long?
> > Can they solve their problems if we just gave them money? We have to pay
> > for dozens of Win95 licenses *anyway* so it's not like it makes a
> > difference to us to give the money to someone else."
>
> Wine and wabi are ultimately irrelevant. When one vendor controls an API
> they simply wait until they day after a new Wine/Wabi comes out add
> 16 new features and use them heavily in their next upgrade to each package
>
> Two months later nothing works again

not only that they add new API calls. They are free to choose which API
side gets ... hmm ... 'slower' or 'not recommended', based on what API
part is used by the competition.

most Microsoft APIs are >largely< redundant. Say i've counted 1500
different API calls in windows.h ... and that doesnt include new bloat
like ActiveX, 3D stuff and the networking nightmare called Netbeui and SMB
over TCP.

Compare this with the 167 Linux system calls.

The Posix API is small, nonredundant and rounded up. Stuff one can build
applications on, which work, will be fast and efficient even say 2 years
later.

just look at the perfectly ok legacy software from the DOS world is being
made impossible in Win32. Does it matter how one opens a file? How one
prints a string or allocates memory?

so 'following' the Microsoft API is suicide from the developers point of
view. It's really a moving target.

-- mingo