LurfysMa said:
OK, you either won't or can't understand its pertinence.
I don't see "logical consistancy" anywhere in those 3
badly-written paragraphs. I also don't see the term "independent",
which is why they are so badly written.
Orthogonal means independent. Period. It comes from geometry
(perpencidulat lines) and force vectors in physics, which are
used to show that forces do not affect velocities that are
orthogonal, ie, perpendicular.
....
FTHOI,
http://dictionary.reference.com/search?q=orthogonal
See the section from the Jargon File.
First off, the mathematical concept ot orthogonality is a
generalization of perpendiculiarity. 'Independence' as you seem to
mean is a derivative concept, not a necessary characteristic of
orthogonality. In the strictest mathematical sense, two members of an
inner product space are orthogonal if their inner product is zero.
Perpendicularity and [linear] independence are special cases.
But the term has been adapted to other fields, such as programming. In
that domain, it means an operation shouldn't cause side effects (part
of the Wikipedia text you dismiss because it doesn't force fit
perpendicularity or mention independence). The implicit rounding in
the Excel formula =1.01-1-0.01 is an example of a side effect. This
contrasts with the lack of implicit rounding in =(1.01-1-0.01).
That these two formulas which most users would expect to produce the
same result don't in fact produce the same result is a consequence of
the inconsistent application of implicit rounding, so the inconsistent
application of this side effect.
But even more fun is the fact, which you can confirm for yourself,
that the formulas
=(1.01-1-0.01)-0
and
=1.01-1-0.01-0
ALSO return 8.67362E-18 rather than zero, same as =1.01-1-0.01.
Call it what you want, but if Excel can return different results for
=1.01-1-0.01 and =1.01-1-0.01-0, it's a mess (a term I don't claim to
have any precise technical definition).