Re: Matlab File Format

[Ted Harding]

On 17-Jun-99 Neil Davey wrote:
> I would think that as the datafile is output of the "program", then it
> would be considered reverse-engineering... here is another section I
> just found in the license:
> 
> "Except as expressly provided by this Agreement, Licensee may not alter
> or modify the Programs without the consent of TMW. In particular,
> Licensse may not alter, adapt, translate or convert "M-Files"
> contained in the Programs in order to use those  files with any
> non-TMW software, nor may the Licensee incorporate or use "M-Files" or
> any other part of the Programs on ir as part of another computer
> program."
> 
> since the data file is an output of the program... you are indirectly
> reverse-engineering the software to workout it's format.. thus I reckon
> breaching the license agreement...
> 
> Neil

This overlaps with a question which is causing great concern in Europe
at the moment (though not to enough people, apparently ... ).

This is the question of the patentability of software and algorithms.

US law, it seems, allows an algorithm or concept to be patented, even if
it has been published so that it has become part of the working knowledge
of many people. European law, at present, does not. There are, however,
signs that under pressure from <guess who> there are clear movements
towards the US position in the European legislature.

Now consider a MatLab M-file. Among the aspects of it, you may consider
that a person, even if they already knew and thoroughly understood the
algorithm, could wish to save development time by simply copying the text
of the file into another software package (with minor modifications) in
order to implement the algorithm that the M-file incorporates. This would
clearly violate the condition quoted above.

Alternatively, a person may wish to acquire knowledge of the algorithm by
studying the M-file (as a substitute, perhaps, for studying textbooks
or published articles). Having done so, that person may then write their
own program file to implement their knowledge of the same algorithm.
In the case of a software package such as octave whose intrinsic structure
closely matches MatLab, it might be very difficult to write such a file
without producing something recognisably similar to the original MatLab
M-file. Nevertheless, provided the coding was done as an "original work",
i.e. produced from scratch but based on algorithmic knowledge acquired
by studying the MatLab file, I reckon that under European law this would
not constitute a breach. (I could be wrong ... ).

Finally, many algorithms and mathematical procedures are common knowledge
yet it might again be difficult to write an octave M-file which was
not recognisably similar to a MatLab M-file which did the same job,
even if the programmer had never seen the MatLab file.

In the words of one biologist on the topic of convergent evolution
(the appearance of very similar features in unrelated species):
"After all, there's really only one way to design a good knee-joint".

Even this could come under restriction if algorithms and concepts
became patentable (independently of the implementation). Lest you
think this is far-fetched, note that at this moment there is
a patent (under application or already granted, I'm not sure which)
for the concept of "style sheet" in document formating software.
Again, <guess who> is involved. Since the style-sheet concept as
described in the patent looks to me as though it covers even things like
the ancient troff macro packages, this leaves me wondering where we're
heading.

But, turning back from this general excursion into "intellectual property
rights", I think the lesson for Octave development and MatLab file formats
is that the whole question is sufficiently precarious for it to be
rigorously excluded from anything that might be considered "official
octave developmnt". Of course some users of octave, in the privacy of
their own dens, may tinker with what they like and share it with
consenting friends. That is their business; but whatever they do
should, as a matter of prudence at least, be kept away from "official
octave".

Ted.

--------------------------------------------------------------------
E-Mail: (Ted Harding) <address@hidden>
Date: 17-Jun-99                                       Time: 10:58:24
------------------------------ XFMail ------------------------------



---------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.  To ensure
that development continues, see www.che.wisc.edu/octave/giftform.html
Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
---------------------------------------------------------------------