Time Histories Status

Log for 12/30/98 **************************

Added time histories for the number of hit per event broken down by sector. I found the following.

1.
It works (sort of) and I have mucked up the web page quite a bit. Sorry. Here is a plot of some results.


\begin{figure} \vskip 3in \epsfxsize 5in \hspace{0in}\epsfbox{SectorHitsPerEvent2.ps} \end{figure}

2.
Part of the reason the web page is so mucked up is that I was getting an error due to the length of the name of the quantity. I originally used `SectorHitsPerEvent' and got the following error at initialization. 
make_db_tables SQL_ERR: parse error near 'DOTOnlineRecsis (tou TIMESTAMP, run INT UNSIGNED NOT NULL, event INT UNSIGNED NO' at line 1
make_db_tables SQL_ERR: query errorCREATE TABLE SectorHitsPerEvent1
I shortened the names to `SnHitsPerEvent' and things worked.

3.
I'm getting lots of broken plot icons for the time histories. I can sometimes twiddle the time range and things will suddenly work. This will have to be made more robust or our ever-persistent collaborators will punt on it. A couple of the plots remained proken no matter how I twiddled the time range. The latest value in the table always looked correct even for the `broken' plots. The histograms used to extract the centroid and the width all look reasonable.

Log for 12/29/98 **************************

As of today online recsis has been modified to pass results to the time histories data base. For example, in online recsis a histogram is filled with the number of hits per event in those events with tracks. An example is shown below.


\begin{figure} \vskip 3in \epsfxsize 3in \hspace{0in}\epsfbox{hitsperevent1.eps} \end{figure}

Notice the centroid is about 296 hits with a width of about 87 hits. This result is roughly consistent with what I see on the web-based interface at JLAB when extracting the most recent table entry (I get the same 290 +/- 90 hits from the table). However, when I go to view a plot of the time history of the centroid (and width) of the hits per event (currently misnamed as hits per track) I get the following results.


\begin{figure} \vskip 3in \epsfxsize 5in \hspace{0in}\epsfbox{HitsPerTrack6.ps} \end{figure}

The points should be roughly constant at 290 +/- 90 hits which is clearly not what the plot shows. The input to this plot had a very wide time range time including an upper time limit a couple if days in the future. In the course of fiddling I narrowed the time range down to something within hours of the known times and get a plot that appears to be correct.


\begin{figure} \vskip 3in \epsfxsize 5in \hspace{0in}\epsfbox{HitsPerTrack5.ps} \end{figure}

It may be that having too large a time range confusing the data base search somehow???

I have also seen the following error message intermittently when I'm running online recsis. The indented sentence is my own debugging message. 

RECS_BRUN  L: Run was created at: 10:13:55 on Mon the 16th of Feb  1998 
update_tables SQL_ERR 0: connection errorBad handshake
 Update db with run   8774 at event   5373 after   11
update_tables SQL_ERR 0: connection errorBad handshake
 Update db with run   8774 at event   12983 after   22


 

1998-12-30