svn commit: r512705 - in /spamassassin/trunk/masses: mk-baseline-results runGA

[jm...@apache.org]

Author: jm
Date: Wed Feb 28 03:18:29 2007
New Revision: 512705

URL: http://svn.apache.org/viewvc?view=rev&rev=512705
Log:
STATISTICS files should include the full freqs report, on the entire training log set, not the 1/10th test set

Modified:
    spamassassin/trunk/masses/mk-baseline-results
    spamassassin/trunk/masses/runGA

Modified: spamassassin/trunk/masses/mk-baseline-results
URL: http://svn.apache.org/viewvc/spamassassin/trunk/masses/mk-baseline-results?view=diff&rev=512705&r1=512704&r2=512705
==============================================================================
--- spamassassin/trunk/masses/mk-baseline-results (original)
+++ spamassassin/trunk/masses/mk-baseline-results Wed Feb 28 03:18:29 2007
@@ -40,7 +40,7 @@
 echo
 # don't just use "freqs", it's often out of date w.r.t. scores
 # remove T_ test rules from the logs
-perl hit-frequencies -x -p -s $SCORESET spam-test.log ham-test.log \
+perl hit-frequencies -x -p -s $SCORESET spam.log ham.log \
 	| egrep -v '  T_'
 
 version=`( cd .. ; ./build/get_version )`

Modified: spamassassin/trunk/masses/runGA
URL: http://svn.apache.org/viewvc/spamassassin/trunk/masses/runGA?view=diff&rev=512705&r1=512704&r2=512705
==============================================================================
--- spamassassin/trunk/masses/runGA (original)
+++ spamassassin/trunk/masses/runGA Wed Feb 28 03:18:29 2007
@@ -133,7 +133,9 @@
 # Statistics generation, once everyone likes the scores
 
 
-  # use the test logs we saved
+  # use the logs we saved
+  fulllogh=$LOGDIR/NSBASE/ham.log
+  fulllogs=$LOGDIR/SPBASE/spam.log
   testlogh=$LOGDIR/NSBASE/ham-test.log
   testlogs=$LOGDIR/SPBASE/spam-test.log
 
@@ -145,6 +147,10 @@
   rm -f ham-test.log spam-test.log
   ln -s $testlogh ham-test.log
   ln -s $testlogs spam-test.log
+
+  rm -f ham.log spam.log
+  ln -s $fulllogh ham.log
+  ln -s $fulllogs spam.log
 
   [ $KILL_SANDBOX_RULES = y ] && rm ../rules/70_sandbox.cf