[Python-Dev] [Fwd: Discussion: Introducing new operators formatrix computation]

Gregory Lielens gregory.lielens@fft.be
Fri, 14 Jul 2000 11:57:37 +0200

Paul Prescod wrote:
> Huaiyu Zhu wrote:
> >
> > ...
> >
> > I wonder how orthogonal would this be against the main development?  Would
> > it require a rebuild for each new release.  Would it also require a patch
> > each time?
> I think that the idea is that you could install new syntaxes in the same
> way you install new modules...of course only after careful
> consideration!!!

It seems potentially more dangerous than statically adding a new
operator, no? If two modules
install incompatible changes of syntax, for exmaple if they demand both
@ as operator, they will be considered as incompatible? On the other
hand, if @ is buil-in, it can be overloaded depending on the type of the
operands, which seems safer...I am not sure, but it seems that if syntax
is module-dependent, things becomes really complex...
> > Look it this way.  Text processing do have their special syntaxes.
> Okay, but probably 90% of Python programmers do text processing at some
> point. Not so for matrix calculation. You believe that the Python world
> would grow greatly if we did matrices natively. I don't know if I
> believe that.

Well, that kind of change could potentially attract all the numeric
community...which is probably a minority (and becoming smaller, based on
the evolution of the number of science students in belgium) compared to
the rest of the comp world, but still a significant minority. At least,
it will attract one person for sure: me :-)
Anyway, it seems that that audience (numeric people) was targeted by
python since the beginning, because why the hell include complex numbers
as built-in type if not? They are of a less general usefullness than
matrices, imho, the last can at least be used for graphic stuff...