[REBOL] Re: RFC: Cross-language benchmark proposal
From: atruter:labyrinth:au at: 8-Nov-2002 17:53
Hi Joel,
My (perhaps overly simplistic) perspective is that I'd want to see a large table with
each column having a different language and each row containing a basic
language construct. If a particular language does not support that construct then the
nearest functional equivalent is used, or a note provided indicating its
absence in the target language.
Beneath each code fragment some quantified benchmark results would be given. Deriving
the "units" of comparison would be the hardest and most politically sensitive
part! ;) Beneath this might be some, perhaps subjective, ways of measuring the code.
A typical entry (formatted as a row) might read:
Construct
foreach
Benchmark
Integer: 1,000,000 integers
String: 1,000,000 strings of length 1-255
REBOL
foreach val values []
Integer Benchmark: n units
String Benchmark: n units
Efficiency: ***
Compactness: ****
Clarity: *****
PERL
foreach $val (@values) {}
Integer Benchmark: n units
String Benchmark: n units
Efficiency: ****
Compactness: ***
Clarity: ****
This would allow folks to see how the fundamental building blocks of each language are
implemented and give a rough guide as to where the strengths and weaknesses
of each are. I remember when first starting with REBOL often saying to myself, "In PERL
I would break this down into steps A, B, C which can be done in Pascal as A,
B or C; or in some completely alien way in REBOL!" ;)
I know that people have to "think different" when using REBOL, but most simple constructs
are present and that is what folks looking at REBOL for the first time and
/ or comparing it to something else will want to see initially.
Regards,
Ashley