[Edu-sig] The fate of vpython

[Dan Crosta]

I don't know much about VPython or NumPy, but I've been involved in  
both contributing to and forking open source projects, from both  
sides, and I would caution against forking a project unless you're  
planning to take it in a new direction, or if there's some other  
compelling reason not to continue contributing from the main branch.

If the author of VPython is not planning to work on it for several  
months and you aren't planning to do anything other than make it  
compatible with a soon-to-be released version of one of its  
dependencies (is that the relationship b/w VPython and NumPy?), then  
why not carry on work from 4.0b5? Surely the dev will welcome any  
contributions which reduce work he'd later have to do anyway. He may  
even be willing to give you write access to the repository if you can  
demonstrate to him your willingness to work hard and ability to rally  
the troops of that class at Cornell.

This way you avoid duplication of work, since the VPython devs will  
neither have to duplicate your compatibility patches nor port your  
patches from 3.x to 4.0. The community is almost always better served  
by contribution than by forking, IMO.

- d


On Oct 6, 2006, at 8:32 AM, Arthur wrote:

> This from the August 10 announcement of VPython 4.0beta5
>
> """
>
> In the "Recent developments" section of vpython.org you can read a
> summary of the new features, which is also included in the  
> documentation
> contained in the release. This includes a description of remaining  
> known
> bugs, some of which are serious, which means that you may enjoy  
> playing
> with the exciting new features, but be aware that there are problems.
>
> I am not going to be able to do any more work on VPython for about  
> three
> months, as the publisher of the physics textbook Ruth Chabay and I  
> have
> written needs a second edition completed by November. Jonathan
> Brandmeyer has graduated from NCSU and is moving on to other  
> things. So
> unless someone else steps forward, it is possible that this version 4
> beta is frozen for several months.
>
> Bruce Sherwood
> """
>
> Essentially it reads to me that the 4.xxx branch of vpython is in  
> some netherworld and, probably, deep trouble, with
> serious bugs, and no developer.
>
> Don't know what Bruce's plan might be after completing his textbook  
> work, but I think he would acknowledge that
> he's not the guy to do the hands-on development work to put things  
> back on track.  Problem is, neither am I.
>
> But I will try to begin to take this bull by the horns, as time  
> allows.
>
> My plan at the moment is to try to fork from the latest stable  
> 3.xxx branch, with the initial goal of
> making it Numpy compatible.  As Numpy approaches final 1.0 release  
> as the standard for array processing in
> Python, it seems to me that a package such as vpython *needs* to  
> work to support it if it is to survive.
> If it  doesn't survive there is a serious hole in the Python  
> educational world, IMO.
>
> Serious educational projects are already suffering from this lack  
> of compatibility.
>
> See "The World is Your Python"
>
> http://www.physics.cornell.edu/~myers/teaching/ComputationalMethods/ 
> python/WorldPy.html
>
>
> related to a Cornell U graduate physics course
>
> """
> Computational Methods for Nonlinear Systems
>
> This course is a graduate computer laboratory, emphasizing hands-on
> programming with a variety of data structures and algorithms. We focus
> on the next generation of tools for computation, simulation, and
> research in a broad range of fields
> """
>
> where vpython is among the required packages, but where they face:
>
> """
> The current version of VPython, however, is built on top of the older
> Numeric package, and seems a bit fussy when used in conjunction with
> the newer NumPy.
> """
>
> "Fussy" I think being a bit of an understatement
>
> I have gotten as far as to set up my ubuntu environment with  
> necessary dev libraries for building against Python2.5
> - boost, gtk2, gtkglarea,etc. and atc. and began to explore  
> the .cpp moudles related to Numeric compatibility.
>
> It would be great to see something concrete come out of all this  
> edu-sig chaos, brainstorming, bickering and worse.
>
> So I would prefer to see this fork of vpython from 3.xxx live as a  
> community project, with *real* C developers involved,
> rather than myself learning as I go and perhaps quickly finding  
> myself over my head and throwing up my hands.
>
> Any interest out there??
>
> Art
>
>
>
>
>
>
>
> _______________________________________________
> Edu-sig mailing list
> Edu-sig at python.org
> http://mail.python.org/mailman/listinfo/edu-sig
>