I just talked to Mike and double checked the method I proposed for skipping
events, and I believe that it will work (and its a little more straight
forward than Mikes event).
(The details are that the Driver's process method is the method that calls
the Driver's Processors, so if the process method is overriden, and
super.process() is not called, the Processors added to the Driver will not
be called for that event).
Tony
----- Original Message -----
From: Michael T. Ronan <[log in to unmask]>
To: Eugene Guillian <[log in to unmask]>
Cc: [log in to unmask] <[log in to unmask]>
Sent: Wednesday, April 21, 1999 11:33 AM
Subject: Re: Skipping events in JAS
> Gene,
>
> I was writing the following reply when I received Tony's response to
> your query. His solution of using a predicate is elegant. However, it
> won't work for you if you've added the TrackReco processes to the
> driver's list. Please also look over the approach that I've outlined
> below.
>
> I use a Reconstruction class to select how many (or which) events I
> analyze. I've modified it to also allow selecting events. I've attached
> the modified version which includes the following elements and methods:
>
> static int FirstEvent=1;
> static int LastEvent=1000000;
>
> static boolean GoodEvent = true;
>
> public static int getFirstEvent() { return FirstEvent; }
> public static int getLastEvent() { return LastEvent; }
> public static void setFirstEvent(int first) { FirstEvent = first; }
> public static void setLastEvent(int last) { LastEvent = last; }
>
> public static boolean isGoodEvent() { return GoodEvent; }
> public static void setGoodEvent(boolean good) { GoodEvent = good; }
>
> I use it in each process with a line like
>
> if
> (nEvt<Reconstruction.getFirstEvent()||nEvt>Reconstruction.getLastEvent())
> return;
>
> However, this version with GoodEvent is not know by the server, and it's
> not easy to fool the server unless you're willing to rebuild its code. I
> would suggest the following.
>
> In your main process, TrackJetAna, execute TrackReco's processes. That
> is, take out the
>
> add(new TrackReco());
>
> line in your driver JetFinderTestV5, and in TrackJetAna's constructor
> and processor methods add
>
> public TrackJetAna()
> {
> // Finds tracks in main tracker.
> tracker = new hep.lcd.recon.tracking.tpc.TPCReco();
> // Add reconstructed tracks to event
> addReconTrks = new AddReconTrks(tracker);
> // Turn off track fitting
> addReconTrks.setNoFitting();
> ...
> }
>
> public void process(LCDEvent event)
> {
> ...
> if (!Reconstruction.isGoodEvent()) return;
> tracker.process(event);
> addReconTrks.process(event);
> ...
> }
> hep.lcd.recon.tracking.Tracker tracker;
> hep.lcd.recon.tracking.AddReconTrks addReconTrks;
>
> It would be more direct to modifiy the base Driver class to check flags
> or whatever, but we'll need to review the design after Sitges.
>
> Mike
>
>
----------------------------------------------------------------------------
----
> /* Reconstruction.java */
>
> public class Reconstruction
> {
> static int FirstEvent=1;
> static int LastEvent=1000000;
>
> static boolean GoodEvent = true;
>
> public static boolean Hist = true;
> public static boolean MCHist = true;
> public static boolean TrkHist = true;
> public static boolean CalHist = true;
> public static boolean PIDHist = true;
> public static boolean EvtHist = true;
> public static boolean AnlHist = true;
>
> public static int getFirstEvent() { return FirstEvent; }
> public static int getLastEvent() { return LastEvent; }
> public static void setFirstEvent(int first) { FirstEvent = first; }
> public static void setLastEvent(int last) { LastEvent = last; }
>
> public static boolean isGoodEvent() { return GoodEvent; }
> public static void setGoodEvent(boolean good) { GoodEvent = good; }
> }
>
