VanLUG Email Archive
> On Sun, 30 Aug 1998, Kaz Kylheku wrote:
>
> > Let's see, there is OpenBS, Net BSD, FreeBSD, BSD/OS... Then there
> > are operating systems that were derived from BSD code like SunOS,
> > ancient HP-UX, ...
Oops, did I really write OpenBS? That was unintentional---I do not
resort to such childish derision.
> If your going to go that far, then remember where linux derived from.
>
> BSD 4.4 and System v3
When you say that Linux is a derived work based on these two, you are making a
concrete claim that is easily refuted, since it was written from scratch. It is
an original work, not a derived work.
You can say that it is based on _ideas_ derived from UNIX predecessors,
and that it largely conforms to their programming interfaces, but that's
something else entirely. These ideas and interfaces are relatively
straightforward to imitate basd on their semantic description, without
borrowing implementation.
On the other hand, SunOS was really derived from Berkeley sources. It is an
actual derivative work in the copyright sense, is it not?
Linux is licensed in such a way that it couldn't become incorporated into
a work that isn't free.
I see this as a major advantage and a powerful incentive to use and back Linux.
It means that whatever work is done on Linux, if distributed, will be
distributed to everyone with source code, otherwise it will breaks the law.
If some company were to distribute a much improved work derived from Linux, it
would steal the wind out of the sails of the mainstream effort, to use a
hackneyed phrase. The free version could come to be seen as something lesser
than the version with the proprietary extensions. In other words, it would be,
in terms of perception, where the various free BSD's are at.
Now you can tell me that some BSD variant does this itty bitty thing better, or
that it's frobozz function is optimized to one machine cycle per iteration, but
these nerd arguments don't amount to anything of significance in the real
world, not to mention that they are transient in the sense that they only have
a basis in the comparison of specific releases of specific software systems.
The truth is that it is the overall performance of a system (in a given
configuration and load situation) that matters. And performance only matters
that much. Companies that standardize on a particular OS platform for their
entire operation rarely do so because the system has optimal performance for
each situation in which it is deployed. Even in the computer geek culture,
optimal performance isn't universally accpted as the greatest virtue, else
nobody would ever have used LISP---but I digress.