Extreme Programming - Re: re-use in physical simulations

Author

Re: re-use in physical simulations

galathaea

2006-10-03, 6:59 pm

Daniel T. wrote:
> galathaea@veawb.coop (galathaea) wrote:
>
> I'm not going to discuss the politically charged questions here, because
> this is not a politically oriented newsgroup.

i would prefer that attitude myself

my only concern is that the political content
may somehow change the technical response

>
> I think reuse across domains has been found to be a chimera. A "Person"
> in a banking app is far different than a "Person" in a video game for
> example.

i do not think this is any different than any reuse

you do a domain analysis
you identify actors with some naming scheme
and look for reusable components that satisfy the roles

obviously some components may be "simply named"
in the same manner as the actors of the analysis
but may not play the indicated roles

a physics vector
would not substitute for a container vector
without confusing many roles
(fixed dimensions versus dynamically varying dimensions
interpretation in spaces with dual products defined
etc.)

identifiers can often be made more specific
through a language's namespacing mechanism

ie.
namespace Banking
{
#include "Person.h"
}

> In the spin you have given your question, you imply that
> "physical simulation" is a single domain when it isn't. A video game
> physical simulation is far different than a weather physical simulation,
> which is different still from a vehicle aerodynamic physical simulation.

i did not mean to imply anything like this

let me try to expand
and please ignore the political data
as only illustrative of the concept in a given domain

a friend of mine wants to do simulations
of the collapse of world trade 7
she has certain non-standard beliefs in this area
but due to her training in physical simulations
she wants to be scientific about it
and run some tests

i usually play devil's advocate to her
but i see such simulations as
potentially offering new engineering solutions
to prevent future collapses

and i have a game physics engine
i have been working on over the years

so
as good americans
we decided to try to find ways to make money
off a national tragedy
by developing some of these simulations
(which she eagerly hopes are actually futile
and goads me on)

now she went and asked the NIST
which has run numerous simulations
of the collapse of the two towers and the nearby wtc7
for a copy of their algorithms and data
(actually, she talked her mate into doing it
who has a more relevant background =) )

the biggest reuse benefit would be the skeleton data for wtc7
as my current models is a very weak mockup from the basic data
found in the FEMA report

very basic composite beam placement
the three main gravity column transfers
the cantilever column transfer on floor 7
four moment frames

but the details can be very important for such models
and it is expected that the data from NIST's models
are more professionally laid out and more detailed

the algorithms used would also be useful
though we have some idea from the NIST report on the two towers
and I suspect my own algorithms
(common statics calculations from a first-year course
with some basic newtonian dynamics)
would give similar results

neither i nor my overeager friend
though
considered it unwise to ask for such details
as we have both done so for published simulations work in the past
(myself in an undergrad paper whose data made it into
work on three-body recombination
and my friend whose grad thesis in optimal circuit design
reused random graph algorithm work of another team)

personally
i think these details have little to do with
what is wise in scientific and engineering simulations
and are more focussed on the issue of reuse in software
and was quite surprised at the quoted out-of-context exchange

but i may be missing some important point here

> So, to answer your question directly: No, but you can't expect much
> reuse across different physical simulation domains.

nor would i expect reuse
based merely on names
without having done a role analysis

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
galathaea: prankster, fablist, magician, liar

Ron Jeffries

2006-10-17, 6:58 pm

On 16 Oct 2006 11:59:34 -0700, "galathaea" <galathaea@gmail.com> wrote:

>although any given file
> or usually any module or subsystem
>should be consistent
>an entire team shouldn't be asked to follow one style
>
>this
> i believe
>can be a hinderance to agility in the long run
>
>engineers are by nature very obsessive
> and tend to become focussed on style issues over substance
>
>this should be avoided
> particularly at the level of codifying into process

If all the programmers code in their own style, won't that get in the way of one
programmer maintaining the code written by another? Won't it get in the way of
programmers doing pair programming?

--
Ron Jeffries
www.XProgramming.com
I'm giving the best advice I have. You get to decide if it's true for you.

Sponsored Links