[Edu-sig] creating an interface vs. using one (Michel Paul)

[Arthur]

Jason Cunliffe wrote:

>Well if you are looking for a good business decision then *please* 
>specify and work to compile an uber-useful LiveCD Edu-Sig distro with 
>all the math-edu-geo goodies...
>  
>
Jason, what edu-sig have you been visiting???

Do you see a group of people here capable of cooperate decisionmaking? 

*My* LiveCD is specified.  Python2.5, as is,  + Numpy + Vpython.  I 
honestly would choose not to confuse the issue further.  All we should 
be thinking about in a distribution of this kind is infrastructure and 
framework and standards - and that, AFIAK, is all we need.

Where is xturtle, where is pygeo, where is crunchy, where is the 
tetrahedron????  Different realm of things, IMO.

Guido in fighting a different strategic battle declared Django to be the 
defacto standard web framework, knowing I am sure he would take fire. 

The Numeric/Numarray/Numpy is already the defacto standard 
multi-dimensional array and linear algebra package.  Because there have 
been cooperate efforts made to keep it a progressive evolution rather 
than a splinter of projects.

I think we need to think here also about defacto standards, rather than 
the kitchen sink of interesting alternatives.  Which  is  why the LiveCD 
of the kind I think you have in mind is not the answer.

I have had absolutely nothing to do with the creation of Vpython.  PyGeo 
uses it because it serves it needs.  It is versalite and serves many 
other kinds of needs as well.  

I think it good strategy to push it out front as a defacto standard for 
basic educational visualization, simulation and modeling tasks.  
Strategic decisions like this necessarily involve a degree of 
compromise, and someone with the authority to make a decision that such 
a compromise is in fact sound strategy.

But here we are talking about a library that is so small and basically 
unpretensious that I see no way of pushing it to the fore without making 
it part of the standard distribution, which itself assumes the 
introduction of some of the Numpy functionality as well, which it is my 
understanding is something already on the drawing board, having Guido's  
attention and implicit support.

Frankly I think Guido is removed from the day-to-day in this realm so  I 
would not expect him to be willing to make a vpython decision.  He only 
made the Django decision when his google work immersed him the the issues.

So I have no answers - only  ideas that are not driven by self-interest 
or self-aggrandizement. And which have some degree of  strategic merit. 
Some, as in worth discussing, at a minimum.

I am paid to advise in the development of business strategies, by people 
who could go elsewhere but who know me well and with whom I have 
developed a track record. And the most important track record I have is  
the ability to separate my self-interest from their interests.  My fee 
is X if the deal happens, Y if it doesn't and X>Y.  I work with people 
who are confident that if I don't like the smell of the deal, they will 
know about it, adamantly.

But they are also people too smart to follow my advice as from a 
mountaintop, and yes I have been  known to be dead wrong - but usually 
within some shooting range and I do a pretty decent job of stating my 
range of certainty up front.

This is in the 8.4 to 8.8 out of 10 range - that something substantial 
will be achieved relative to Python's acceptance in educational 
settings.. But what is the risk associated with being wrong???  By my 
analysis, nothing terrible. An extra library.  So we have a go, basically.

What I can't judge are the depth of the technical difficulties.  My 
intuition that they are  manageable is nothing more than intuition  - 
especially when we are discussing the  more exotic platforms.
 
Art

















>That way you avoid conflicts and friction. This distro is free to 
>promote, develop and demonstrate a __range__ of tools covering a range 
>of tastes, philosophies, needs and competences on one disk [package, 
>torrent, ISO, pick your medium].
>
>  
>


>Install tested and run on existing 'puters [WinXp etc without any 
>collateral damage]
>It can include brave new 21st century alternative OS curriculae.
>Then When If it is a obvious success, don't be surprised that the 
>default standard distribution sensibly absorbs the relevant parts someday.
>Pioneers always lead administrations and institutions right?
>
>Jason
>_______________________________________________
>Edu-sig mailing list
>Edu-sig at python.org
>http://mail.python.org/mailman/listinfo/edu-sig
>
>
>  
>