svn commit: r572167 - /spamassassin/trunk/build/automc/freqsd

[jm...@apache.org]

Author: jm
Date: Sun Sep  2 14:34:07 2007
New Revision: 572167

URL: http://svn.apache.org/viewvc?rev=572167&view=rev
Log:
revert to using old corpus-hourly; the new script still has bugs, breaking '.all' report generation for the ruleqa system, and I don't have the time to fix them just yet

Modified:
    spamassassin/trunk/build/automc/freqsd

Modified: spamassassin/trunk/build/automc/freqsd
URL: http://svn.apache.org/viewvc/spamassassin/trunk/build/automc/freqsd?rev=572167&r1=572166&r2=572167&view=diff
==============================================================================
--- spamassassin/trunk/build/automc/freqsd (original)
+++ spamassassin/trunk/build/automc/freqsd Sun Sep  2 14:34:07 2007
@@ -33,8 +33,8 @@
 # by using IPC::DirQueue,
 
 # the parent process continually generates the faster reports
-my $parent_reports = "DETAILS.new DETAILS.all DETAILS.age ".
-		"NET.new NET.all NET.age SCOREMAP.new";
+my $parent_reports = "LOGS.all DETAILS.new DETAILS.all DETAILS.age ".
+  "NET.new NET.all NET.age SCOREMAP.new";
 
 # the child process generates the slow reports
 my $child_reports = "OVERLAP.new";
@@ -123,22 +123,15 @@
 
   my $job = $dq_fast->wait_for_queued_job($this_sleep);
 
-  # the "fast" jobs are in charge of copying the logs to the data
-  # storage area
-
   if ($job && $job->{metadata}->{dir}) {
     # if a dir was specified, it's always a "b" (buildbot) mass-check;
     # that's the assumption here at least
     logit "starting buildbot-requested faster reports";
-    run ("cd masses/rule-qa; ./mk-ruleqa-reports ".
-          "--reports='$parent_reports' ".
-          "--tag=b --copylogs --dir ".$job->{metadata}->{dir});
+    run_corpus_hourly($parent_reports, "--tag=b --dir ".$job->{metadata}->{dir});
   }
   else {
     logit "starting rsync faster reports";
-    run ("cd masses/rule-qa; ./mk-ruleqa-reports ".
-          "--reports='$parent_reports' ".
-          "--tag=n --copylogs");
+    run_corpus_hourly($parent_reports, "--tag=n");
     # may also be weekly.  no way to differentiate currently until
     # AFTER corpus.hourly is run!  TODO?
   }
@@ -153,22 +146,17 @@
 # ---------------------------------------------------------------------------
 
 sub child_loop {
-  my $job = $dq_slow->wait_for_queued_job($idle_sleep);
-
-  # create slow reports.  note that this uses the logs copied by
-  # the "fast" freqsd thread
+  my $job = $dq_fast->wait_for_queued_job($idle_sleep);
 
+  # add switches
   if ($job && $job->{metadata}->{dir}) {
     logit "starting buildbot-requested slow reports";
-    run ("cd masses/rule-qa; ./mk-ruleqa-reports ".
-          "--reports='$child_reports' --copylogs ".
-          "--tag=b --dir ".$job->{metadata}->{dir});
+    run_corpus_hourly($child_reports, "--tag=b --dir ".$job->{metadata}->{dir});
   }
   else {
+    # create slow reports
     logit "starting rsync slow reports";
-    run ("cd masses/rule-qa; ./mk-ruleqa-reports ".
-          "--reports='$child_reports' ".
-          "--tag=n");
+    run_corpus_hourly($child_reports, "--tag=n");
 
     # recreate the corpus link-farm
     logit "running 'freqsd-infrequent' tasks";
@@ -178,6 +166,13 @@
   logit "completed slow reports";
 
   if ($job) { $job->finish(); }
+}
+
+sub run_corpus_hourly {
+  my ($reports, $opts) = @_;
+  run ("cd masses/rule-qa ; ./corpus-hourly ".
+        "--override='output_classes=$reports' ".
+        "$opts");
 }
 
 # ---------------------------------------------------------------------------