Hi,
after a few phone conversation with Jason and Eugine, I looked into the
warnings dumped by lrn.
turned out the ABS:STATUS.B# channels read from coda file (not directly
from epics channels) are all 0 . I can caget ABS:STATUS.B1 to
ABS:STATUS.BC and get the right bits but when lrn reads the same channels
from the coda file, they are all 0.
I am not sure what is the cause. But the good thing is that epics target
bits are not as critical as scaler information to the crunching. I
following the practice in charge.C to use scaler target bits to control
charge integration in lrn. So scaler target defined and scaler target
states are crucial to the correct integration of charge on all the states.
I read also epics information but they are only read, compared to scaler
data if the same piece of information is registered in both epics and
scaler. Warning message are produced when there are discrepencies. so
although a lot of warnings have been made on epics target undefined and
so on, data crunching should be fine.
I have removed the warning about epics target data from lrn.
Chi
On Fri, 13 Feb 2004, Bill Franklin wrote:
> Hi Chi,
>
> A lot of warnings about the target state were being printed as last
> night's runs were crunching. However, the ABS EPICS displays indicated
> that the target was consistently in a well-defined state (1) and the
> target parameters were stable. So it's not clear why the software
> wasn't reading a well-defined target state.
>
> Does the online GUI run checks on the target state? It's still unclear
> why it is not successfully reconstructing any events (the momentum,
> vertex, and angular distribution histograms are still empty).
>
> Bill
>
> On Fri, 13 Feb 2004, Chi Zhang wrote:
>
> >
> > Hi, just a reminder to shift persons about data crunching:
> >
> > the new lrn processes epics and scaler data and if it encounters any
> > problem(undefined target, inconsistancies between scaler and epics data)
> > warning messages will be printed to the screen.
> >
> > This typically happens in the first minute of a crunch since epics and
> > scaler in coda file are read through in a independent loop over entire
> > data file before rewinding back and proceeding to physics events.
> >
> > So it may be good if you sit for another 15 seconds after typing lrn, and
> > see if any warning is appeared.
> >
> > Also, the codes are setup to be able to crunch unfinished runs.
> >
> > Chi
> >
>
> --
> **********************************************************************
> Wilbur A. Franklin * Phone: 617-253-9518
> Bates Linear Accelerator Center * FAX: 617-253-9599
> P.O. Box 846 * Email: wafrankl@mit.edu
> Middleton, MA 01949 *
> **********************************************************************
>
>
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:30 EST