Re: [Visualpython-users] strange math

[Bruce S. <Bru...@nc...>]

Right on all counts. I shouldn't try being analytical when I've got the flu!

Bruce Sherwood

Craig Struble wrote:
> Hi Bruce,
> 
> Ah, you're getting into the strangeness of objects. The expression 3*A 
> creates a new object and assigns it to the name A you're using. That 
> won't be the same object assigned to axis.
> 
> In order to do the updates, assuming vectors are mutable, the code 
> should be
> 
> A.x = 3*A.x
> A.y = 3*A.y
> A.z = 3*A.z
> 
> I tried this, however, and it also didn't work because of axis=A 
> behaving like axis=vector(A); that is, arrow creates a copy of the 
> passed in object.
> 
> Creating a copy of the vector is a design decision that's appropriate 
> for the intended audience of VPython. Doing it the other way, using only 
> the object reference, would cause all kinds of side effects that would 
> be difficult to debug.
> 
> This is, however, inconsistent with the general behavior of Python. If I 
> wrote the code
> 
> A = vector(0,1,0)
> b = arrow(pos=(0,0,0))
> c = arrow(pos=(0,1,0))
> b.axis = A
> c.axis = A
> A.x = 1
> 
> As a Python programmer, I expect b and c to use the same vector object 
> for their axis, and changes to that object (A.x = 1), should update the 
> axis for both arrows referred to by b and c. If I intended to treat b 
> and c differently, but to start with the same axis, I would explicitly 
> write
> 
> b.axis = vector(A)
> c.axis = vector(A)
> 
> This all gets back to the intended audience, but making it clear where 
> VPython might not behave exactly like Python in all cases. I can work 
> around each approach, I just need to know my limits.
> 
>     Craig
> 
> On Sep 14, 2009, at 10:38 AM, Bruce Sherwood wrote:
> 
>> Thanks for the additional background, Craig. It's good to have a 
>> card-carrying
>> computer scientist commenting on the issues!
>>
>> Your point about updating a bunch of Visual objects however doesn't work,
>> because the setters for Visual vector attributes like pos or axis make 
>> new
>> vectors. This explains for example why you can write as a shorthand 
>> pos=(1,0,0)
>> instead of pos=vector(1,0,0); a vector is constructed out of the triple.
>>
>> Here is a routine that shows the (non)effect:
>>
>> A = vector(0,1,0)
>> arrow(pos=(0,0,0), axis=A)
>> arrow(pos=(1,0,0), axis=A)
>> scene.mouse.getclick()
>> A = 3*A
>>
>> After clicking, the arrows don't get longer, because axis=A is basically
>> axis=vector(A).
>>
>> In any case, it would be a rare circumstance where you wanted some vector
>> attribute of lots of objects to be the same, and to change together.
>>
>> Bruce Sherwood
>>
>> Craig Struble wrote:
>>> It should be pointed out that many object oriented languages share
>>> this behavior. Java behaves the same way, for example. Many others
>>> have pointed out the difficulties in copying objects (deep vs. shallow
>>> copies) and bigger problems for copying arise when there are circular
>>> references, which are common for complex data structures (e.g. general
>>> graphs).
>>>
>>> The object reference behavior is useful when implementing some kinds
>>> of data structures when you need a "trailing pointer" for updates.
>>> Lots of linked list implementations use this approach. It's atypical
>>> to do this in Python since it has a very rich collection of data
>>> structures already implemented, but it can be useful.
>>>
>>> I would guess in VPython, this behavior could make it easier to update
>>> the positions or movements of a large group of visual objects. If they
>>> all refer to the same point or vector, change that vector and all the
>>> objects change accordingly. It would be nice to make a single update
>>> instead of having to looping over each object, which would probably be
>>> slower.
>>>
>>>     Craig
>>>
>>
>> ------------------------------------------------------------------------------ 
>>
>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 
>> 30-Day
>> trial. Simplify your report design, integration and deployment - and 
>> focus on
>> what you do best, core application coding. Discover what's new with
>> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
>> _______________________________________________
>> Visualpython-users mailing list
>> Vis...@li...
>> https://lists.sourceforge.net/lists/listinfo/visualpython-users
> 
> -- 
> Craig A. Struble, Ph.D. | 369 Cudahy Hall  | Marquette University
> Associate Professor of Computer Science    | (414)288-3783
> Director, Master of Bioinformatics Program | (414)288-5472 (fax)
> http://www.mscs.mu.edu/~cstruble | cra...@ma...
> 
> 
>