On Feb 12, 2011, at 3:38 PM, Isaac Gouy wrote: > On Feb 12, 1:28 pm, Michael Gardner <gardne...@gmail.com> wrote: >> That's why this kind of competition is not interesting to me. As it only >> compares the fastest programs, there's every incentive to submit >> horrifically complex, optimized-to-the-hilt solutions that would almost >> never get used in the real world. >> >> Rather than ask "what's the fastest this can be done in language X?", we >> should ask "how fast are the idiomatic ways of doing this in language X" and >> possibly "how hard is it to do it faster, when those are not good enough?". > > > Looking at the benchmarks game web page showing the "fasta" programs: > - 4 different Clojure programs are shown > - 2 different Java programs are shown > - 4 different C programs are shown > etc > > > http://shootout.alioth.debian.org/u32/performance.php?test=fasta > > >> Possibly just including code size/complexity with the performance metric >> would do the trick, though measuring that cross-language is a challenge all >> its own. > > There is a column showing compressed source code size for each > program. > > > And then there's > http://shootout.alioth.debian.org/u32/code-used-time-used-shapes.php
Those are all good, but how many people actually look at anything beyond the fastest implementation for a given language? To some extent this is their problem, sure, but it's not helped by the prominence the shootout site gives to the fastest implementations in various places. In the first place, I'm not even sure what the purpose of all that data is. What do people do with it? Is it purely for entertainment, or do people actually use it to decide what language to use for a project? If it's just for entertainment (as the word "game" seems to imply), won't people just skim the numbers for the fastest programs and walk away with wrong impressions? In my opinion, it's simply not meaningful nor interesting to compare the fastest implementations across different languages. But by presenting those comparisons prominently (even when it also provides other more meaningful comparisons elsewhere), the shootout site does disservice to its users and the programming community by providing a convenient yet misleading focus point. Perhaps each language could have one "best" implementation selected by the community or by a panel of judges; then the default or most prominent comparisons could use these "best" implementations rather than the fastest ones. Either way, it would be good to remove those fastest-vs-fastest comparisons altogether, or at least reduce their prominence significantly. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en