Re: ADE_Splus

From: Agustin Lobo (
Date: Tue Jan 26 1999 - 18:36:52 MET

(you can write in French, I actually understand it and speak it
better than Eng. It's just the hypercomplicated French ortograph
that prevents me to write it as fast as Eng.)

Total disagreement, nevertheless it's obviously up to you
so I will not go on with the topic beyond this reply
(probably most ADE users just don't care about this question)

On Tue, 26 Jan 1999, Jean Thioulouse wrote:

> This would be easy for computational modules, and probably more difficult for
> graphical modules. The problem is : what would be the benefit ?
Anybody wanting to integrate all its statistical tools within a coherent

> ADE-4 and Splus are not intended for the same users. Splus is a professional
> statistical software, very expensive and not very easy to use for biologists.
> ADE-4 is free, dedicated to multivariate analysis and aimed at ecologists.
> Ecologists will not buy $plus to use ADE-4, and professional statisticians
> just need a single value decomposition function to do what they want in
> multivariate analysis.

First, biologists and/or ecologists are not more stupid than the average
scientist and are perfectely able to learn Splus at the same rate. I
actually might
be an example of an stupid ecologist using Splus since almost 10 yrs ago...

Second, Splus is not difficult, it's actually easier than ADE. It's just
more complex, because it's got hundreds of functions. But everything is
included in a coherent high-level language, so it's not difficult.

Third, you CAN do "everything" with SVA, but the point is not to repeat
what is already done. For example, you can write a function for PCA in
Splus (actually a good and simple exercise for students) but there is
already one called prcomp. The same would hold for other, more complex,
multivariate analysis: you could program them in Splus, but it would
take me a long, long time. Instead, wraping your C objects within
a Splus function would be easier. And instead of producing files and files
for each analysis, we would get one single Splus object: a list, in which
each component would be what we have now in each file. And this components
with be easy to further analyze with the rest of Splus tools.

Fourth, yes, Splus is damned expensive (because it's damned good). This is
why I also suggest XploRe, which seems very simmilar and is cheap
(for Win and Unix) or free (for Linux, ooops! is Linux too difficult
for ecologists?).

Well, I'll keep on exporting and importing from Splus to ADE and
from ADE to Splus...

> OK, this is a little caricatural, and it would not be so easy, as we would have
> to deal with Splus objects structure.

Much easier than C, Fortran etc.


This archive was generated by hypermail 2b30 : Sat Feb 10 2001 - 10:35:55 MET