Re: Troll-o-Matic

Bill Atkins <NOatkinwSPAM@xxxxxxx> writes:

Jon Harrop <usenet@xxxxxxxxxxxxxx> writes:

[...]

Anyway, the ray tracer is only meant to be a little fun.

Is it really? I first came to Lisp last May, and even then you were
filling comp.lang.lisp/comp.lang.scheme with these silly benchmarks
and even then people were telling you your code length measurements
were bunk. What's the deal, Jon Harrop?


I think Jon Harrop gets more flak that he deserves. Quite a lot of the
criticism that was levied against his benchmarks was justified, but Jon also
did eventually incorporate quite a lot of the constructive criticism.

The task is neither so simple as to be entirely without interest nor so
complex as to be completely unintelligible.

And whilst it is certainly true that only quite limited conclusions about the
suitability and performance of various languages for real world tasks can be
drawn from a small toy example like this raytracer, I think those who conclude
that this benchmark is consequently completely useless are overshooting the
mark (it's not like vastly better data of this kind would be widely available
or idiomatically implementing the specific task one is actually interested in
in all these languages for comparison would be practical in most situations).

If nothing this page serves as a nice example of how to optimize a certain
type of code in various languages and to what extent clarity and conciseness
are traded off against performance in the process. I'm looking forward to
seeing further improvements to this benchmark and similar benchmarks for tasks
that stress different language features in the future.

And yes, consecutive whitespace should not play a role in verbosity measures.

'as

It is easy to lie with statistics, but it's a lot easier to lie without them.

-- Richard J. Herrnstein
.

Relevant Pages

  • Re: Statically AND Dynamically Typed Language ??
    ... thrash Lisp at this and almost all other benchmarks. ... I wouldn't say 'thrash'. ... There are many problems with the shootout: ... solved in languages like OCaml and Lisp are simply not represented. ...
    (comp.lang.misc)
  • Re: Statically AND Dynamically Typed Language ??
    ... that thrash Lisp at this and almost all other benchmarks. ... are easily solved in languages like OCaml and Lisp are simply not ... Haskell is a specific language. ...
    (comp.lang.misc)
  • Re: Statically AND Dynamically Typed Language ??
    ... that thrash Lisp at this and almost all other benchmarks. ... solved in languages like OCaml and Lisp are simply not represented. ... Haskell is a specific language. ... GHC program - Haskell doesn't seem to be "thrashing all other ...
    (comp.lang.misc)
  • Re: New Computer Language Shootout?
    ... Complex open-ended problems (like Sudoku) admitting different ... along the way are *bad choices* for benchmarks. ... benchmarking *languages* - not algorithms or programmers. ...
    (comp.lang.lisp)
  • Re: verbosity vs. inscrutability
    ... Jon Harrop wrote: ... > variety of different tasks designed to show the strengths of different ... > types of language, let people implement them in different languages, ... benchmarks drive the things that people bother to observe. ...
    (comp.lang.lisp)