[Shootout-list] scoring discrepancy
Stephen Weeks
sweeks@sweeks.com
Wed, 7 Jul 2004 10:12:23 -0700
Aha! I figured out the problem. The .craps.table at the web site
http://shootout.alioth.debian.org/.craps.table
is not the same as the .craps.table in the CVS. The one at the web
site has CVS merge lines in it, and hence lots of duplicates of
scoring entries. If I run my scoring program on it, I get
clean 73.4552 ~14
gcc 58.8543 ~2
mlton 44.7229 ~2
g++ 43.7682 ~2
cmucl 42.8436 ~25
sbcl 42.1749 ~21
ocaml 40.4054 ~2
gcj 34.7782 ~18
nice 31.1874 ~23
bigloo 30.5982 ~1
These scores agree exactly with what the web page says. The negative
numbers (in SML speak :-) indicate how many tests beyond 25 have been
included.