[Shootout-list] Ray tracer

[José Bollo]

Le dimanche 19 Juin 2005 15:11, Jon Harrop a écrit :
> On Sunday 19 June 2005 13:26, Bengt Kleberg wrote:
> > On 2005-06-19 13:28, Jon Harrop wrote:
> > if the description is imprecise it could be improved upon.
>
> Giving a program is the easiest way to get a "formal" specification.
>
> > if the mapping for all tests is (only) a program, what language(s?) do
> > you suggest we use?
>
> C++ and OCaml works well for the ray tracer. The OCaml might be easier to
> read (anyone?). I've never programmed in Java, C# and haven't used Fortran
> for a long time but I can still read them. So I don't really think it
> matters. IMHO, it will be more compelling for people reading the shootout
> if there is a succinct program (like my 57-line ray tracer). Reading 471
> lines of Eiffel isn't quite so appetising. :-)

You are right my eiffel program for the raytracer is 471 lines.
I understand you.
I can also say that reading the ML and ocaml scripts is very easy ;^|
I coulded make it much compact but i used the standard formater.
So for me LOC does not care, i let people juge by themselves and i respect 
your jugement.

Can i contribute to the debate about raytracer?

If such a test is sensitive to the order of iterations on the scene database, 
then it must be or removed or the order must be strictly explained. 

One has thrown the wal and has applied a new algoritm to comput the shadow. So 
now all contributions have to be adapted to the new bench "howto". In such 
case, can send a mail to the contributors or mark an history to the bench. 
Maybe something is missing. Maybe raytracer went out from sandbox to early.

When you will converge to the final test can you send an official announcement 
on that mailing list?

These are just ideas ;^)

best wishes
josé