The following is a list of bugs in the L2 code, and their solutions, if we've found them.  Apologies for the sketchy page, it's very late at night for me.

1)  The Gamma trigger abort rate goes down as the threshold parameter is increased.  We would expect the opposite behavior, since threshold=99999 should equate to abort all, and threshold=0 should equate to accept all.
----Not resolved

2) The Event counter in the GUI and in Jan's histograms don't agree, though they are close.
----The way the TCU works we may have slightly more or slightly fewer events than originally requested in the GUI, and this number may also differ slightly from the number recorded in Jan's histograms.  This is entirely expected and there is no reason for concern.

3) initRun in the jet algorithm did not return a negative value on fail.  rc_handler expects negative return values if the function didn't complete successfully.
----There was a missing minus sign in the return statements for the jet algorithm as well as the two gamma algorithms.  all three have been corrected so that initRun now returns values less than zero on fail.

4) After 14:00 on Feb 10 the l2gamma online monitors stopped reporting correctly filling the high tower histogram.  The histograms showed events, still.  There should be no way to have listed entries, zero overflow and underflow, and sitll have nothing plotted.
----The failure was traced back to unsuccessful initRun() calls as in (3), due to bad parameters set in the GUI.  To catch this sort of failure in the future (it shouldn't reoccur), the histogram titles were set to CONTACT OPERATOR and INIT FAILED, and are changed later on in the program.  If the alogrithm dies in init, they will show these titles.  However, the histograms should still have plotted something.

4a) Even in the failure seen above, the histograms should have filled correctly.
----This was found to be a bin problem in some places -- bin 100 was being filled but the histogram only went up to bin 99.

4aa) In the problem in 4a, we should report entries that went into bin max_bin+1 as overflow.
----This is a minor problem and can be fixed later.

5) decision() and compute() seem to be called more than once for some events?  Both of these should only be called once per event, per algorithm.

...
If you have any more L2 errors to report, please comment, bearing in mind that this is meant only to deal with gamma, jet, and related spin algorithms.