[Shootout-list] Weekly Update

[Brent Fulgham]

Here's a brief update of the latest Shootout activities:

1.  I have revised the build scripts so that the runlog is now generated=
 with any program error
attached to the bottom.  This log can be accessed on the language page (=
the page that
shows the various tests and their pass/fail state) under a new column ca=
lled "log".  The "log"
is the runlog, and the "out" is the generated output.  This is sometimes=
 useful, when the generated
output does not match for some reason.

Example 1:  Elastic Ackermann test.

The "out" link is a dead link, but the "runlog" link shows:

************************************************************
*   TEST KILLED BECAUSE IT EXCEEDED --maxtime 600 SECS
************************************************************

Example 2:  MzScheme failing Random test

"Out" shows:
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
28.785150891632373

"runlog" shows:
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
COMMAND LINE (%A is single numeric argument):

random.mzscheme_run %A
************************************************************
*   TEST (random.mzscheme) FAILED:  random.mzscheme_out differs from Out=
put.9000
************************************************************
-rw-r--r--  1 brent brent 13 Sep 12 10:04 Output.9000
-rw-r--r--  1 brent brent 19 Sep 19 15:59 random.mzscheme_out

By examining the expected output file, I can see that the problem is MzS=
cheme is
not rounding the number properly.

The logs are in place for everything, but there are a lot of tests that =
I manually
disabled long ago.  I'm trying to correct these where I find them, but i=
f you run
across any dead "runlog" links, please let me know so I can add them to =
my list.

2.  "Chicken" Scheme added:  New language implementation.  Many thanks t=
o Felix
(Chicken's "father") for providing them.

3.  "newLISP" added:  New language implementation.  Many thanks to Lutz =
Mueller
(newLISP's "father") for providing these implementations.

4.  A couple of new tests (I may have mentioned them previously).  They =
are aimed at
showing off some of Erlang's strengths by requiring lots of message pass=
ing (in the
case of the new "ringmsg" test) and loading of plugins (in the case of t=
he new "plugin"
test).  Note that "plugins" is being revised slightly, so please don't b=
other implementing
anything until further notice.

5.  I changed the Matrix test to use =C5smund =D8deg=E5rd's NumPy implem=
entation.  What a
huge speed increase!  Compare 0.39 to 16.22!  (One snide side-note:  new=
LISP does this
at a zippy 0.21!)

TODO:

There are a few items on my todo list.

1.  I owe John Max Skaller an apology for losing his Felix tests in a ma=
iler screw-up.  I've posted
to him off-list apologizing, and asking him to resend them.  This is an =
interesting new language,
and it will be fun to see where it lands in the rankings.

2.  I installed the portable dotNET and dotGNU stuff, so we may be able =
to get some C# timings soon.
If anyone cares to try dotGNU out to see if it suffers from the same pro=
blems as Mono, I would
love to hear from you.

3.  I intend on adding a test to highlight the logic-based languages' st=
rengths (possibly an N-Queens
or some other backtracking problem).  This won't happen for another mont=
h or two when I finish
reading a few logic language books for ideas.  Any logicians with Prolog=
, Mercury, or LISP skills with
devious ideas, please contact me.

4.  I'd like to provide a couple of Scorecard pages:  the current "overa=
ll" page, a "compiler" language page
(for the various optimized, compiled languages), and a "scripting langua=
ge" page for the
Perls and Pythons of the world.  Does this sound useful?

5.  I've got to look into incorporating Isaac's website ideas.  I think =
a lot of his points are valid,
but I cringe at the time I think it will take to go through and revamp t=
he site.  The site is currently
based on the open-source Tigris.org CSS and icons.  If anyone has an exa=
mple of a site with
better readability (etc) we can look at doing that.  In the meantime, th=
e site will stay as-is until
I can get around to modifying it further.  I'd like to hear other people=
's thoughts on the site
layout and design before changing anything.

6.  Very soon I will correct the timing resolution, so it shows the corr=
ect 0.01 resolution.

That's about it.  If you sent me implementations or ideas in the last tw=
o weeks and have not
seen it mentioned here (or in private mail from me) please re-send it.  =
It probably got eaten
in my mailer screw-up!  :-)

Thanks,

-Brent