[Shootout-list] Missing programs and score calculation (Was: Re:TinyCC and Io)

[Jon Harrop]

On Saturday 17 September 2005 19:22, Isaac Gouy wrote:
> The programs you bundled together as ray tracer produce the same output
> by different means

All of the benchmarks on your shootout "produce the same output by different 
means". My ray tracer is no exception. For example:

Ackermann is implemented completely differently by C and SML/NJ (recursion vs 
continuation).

Pidigits uses different libraries for arbitrary-precision arithmetic (e.g. 
OCaml's Bignum vs GMP).

> - which is your idea of what a test should be but is 
> not the Shootout idea of what a test should be.

As the shootout's "idea of what a test should be" is ill defined, that is a 
meaningless statement.

> You know that. 

I know that the versions that have been left on the disembodied shootout page 
were either the fastest under 100LOC or the shortest over 100LOC, following 
my idea. In the face of more disapproval than approval, I was more than 
willing to change them back. Then I found I couldn't change them because I'd 
lost CVS access so they've remained in this undersireable state ever since 
and are now inaccessible from the main pages.

If "the shootout" would rather the ray tracer implementations all used 
seemingly-equivalent algorithms (e.g. the first versions given on my site) 
then I'm happy to change them back.

I'll note again that this shootout requirement is subjective and, 
consequently, is impossible to satisfy. I am willing to try though.

> The programs you bundled together as ray tracer might all have been
> authored by you -

The ones that I put my name on were written by me. There are several other 
versions by other people.

> otoh in various newsgroup discussions you have 
> acknowledged not giving credit to the real authors of some ray tracer
> programs.

No. I have certainly not claimed that other people's implementations were 
written by me.

I acknowledged that (small) parts of my programs were based 
on other people's code. For example, my original SML version included:

  infix 1 +| fun (x1, y1, z1) +| (x2, y2, z2) : vec = (x1+x2, y1+y2, z1+z2)

Matthias Blume then wrote his own SML version that included:

  infix 6 |+| fun (x1, y1, z1) +| (x2, y2, z2) : vec = (x1+x2, y1+y2, z1+z2)

So I copied his use of the precedence "6" instead of "1" without 
acknowledgement. Alex Goldman claimed that this constituted plagiarism. My 
response was "if anyone wants acknowledgement for any contributions that I 
have ignored then please tell me and I will add an attribution to my 
website". That still stands. I do not believe I have included anything 
significant enough to warrant attribution and, apparently, neither do any of 
the original authors as none have contacted me or replied to Alex's post.

> Shootout can only accept programs from the original author - you know that.

Absolutely. People have tried to submit versions in other languages for the 
shootout to me. Following your wishes, I told them to submit the code 
directly to the shootout. Their programs have yet to be appear.

> You repeatedly, rudely and wrongly claimed that programs you had
> contributed to Shootout were being ignored - when in fact those
> programs had never been contributed to Shootout, so it had never been
> possible for anyone to add them to CVS. You know that too.

Unless you have removed them, the ignored contributions will still be on the 
shootout page that I linked to last time you tried to claim otherwise.

Ultimately, if you reinstate my CVS access then I'm more than happy to update 
the ray tracer versions to be as algorithmically equivalent as I can 
reasonably be expected to make them.

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
Objective CAML for Scientists
http://www.ffconsultancy.com/products/ocaml_for_scientists