Hi, Sho.

Thanks for the details.  Responding to your points below…

(hopefully it formats okay!)

> I really insist that we don't make it necessary for shifters to interpret plots.

Some amount of this is inevitable, isn’t it?  Even if there is an automated process to compare against
reference plots, shifters would still need to know how to interpret the comparison.  Is there some
interface other than plots which you would like to see?  I don’t really see that being any easier to use
than simply looking at an output plot and knowing what it should look like.  We can provide detailed
documentation there if necessary.  Or did you want like a list of numbers that characterize some
variables like the fit results?  Did you want a separate panel for it?  To me, this is probably piling too much
functionality onto one application.  The monitoring app is currently designed to show plots from the
LCIO data and give basic information about event rates, etc.  It is not really intended to do much 
interpretation or transformation of those plots.  Does that make sense?  I can see the value in what
you’re suggesting.  I’m just not sure the existing app is the right place for it.

> As much as possible, monitoring app components should be doing that 
> (e.g. fitting plots and comparing to nominal values). There should be a 
> standard way for monitoring app components to report those results, and 
> that presumably involves some software interface that monitoring app 
> components implement, and some GUI code in the monitoring app. Those are 
> the "define interfaces" and "shifter interface" items respectively.

Okay, that’s more clear now, thanks.  Couldn’t this be done through the current system
by implementing this behavior in Drivers?  AIDA can do histogram overlays as well as fitting.
One could also write Driver’s that have input arguments specifying paths to reference plots
in external AIDA files, for instance.  Anything additional in that area will likely require some
pretty major structural changes and additions to the app, which I would rather avoid.  Can we
try to work with what we have for now?

> Some amount of subsystem monitoring is going to have to be done in the 
> monitoring app - for example, SVT alignment should be monitored using 
> online tracking.

Yes, I agree.  I didn’t mean that no sub-system monitoring will be done, as this is kind of
the point of monitoring app, to look at data from the SVT and ECAL.  All I meant was that 
the monitoring app should only deal with data that makes it into the final LCIO events.  
It is not designed to peak directly into upstream data like the EVIO files.  In the case that 
there is some value from the EVIO which must be monitored, then it should be written into 
the LCIO event via the builder/converter class.

> Wired event display is an online display. I'm not the one who came up with 
> that requirement; the people who think it's important (I don't remember 
> who) should clarify.

Thanks.  The requirements here will determine the work load, and I think just the basics of hooking
Wired into the ET system and converting to LCIO is a lot of work.  It will likely require authoring a JAS3 plugin.
I’m not sure I want to take on this responsibility even though I’m listed here.  It should be discussed.  Perhaps
this could be a good project for a new person who wants to learn the framework?

> It makes sense to break down the conditions system tasks. So far I'm 
> hearing "use new conditions system in existing SVT software" "use new 
> conditions system in existing ECal software" and "set up conditions 
> database at JLab." Can you identify people who will take charge of each 
> task? (I assume Omar and I are the natural choices for the first two.) 
> Anything else?

Yes.  I agree with adding those new items.  I can setup the database at JLab but I’ll probably need help
from some JLab contacts to find out all the details.  Likely I would need to work with the sub-system experts
to load some of the basic conditions like channel and DAQ maps correctly.  So there will actually be a number of
people involved with it, but you can list me for that one.

> 
> I'm opening this up to hps-software because I think we need a general 
> discussion of monitoring before we can start on some of these items.

Agreed! 

> 
> On Wed, 30 Oct 2013, McCormick, Jeremy I. wrote:
> 
>> Hey Sho.
>> 
>> Need some clarification on schedule items.
>> 
>> "define interfaces for monitoring app components?
>> 
>> What does this mean?  The basic interfaces there are defined.  One writes lcsim Drivers to produce the plots based on data from the event.  If there is something more that is needed then let me know specifically.  I?m not sure we want to add any more major functionality to the existing app so if there are new features that are desirable we should think about either trying to use the existing system or making separate applications.  I know that one major feature I would like to add is trending plots, e.g. ?strip chart? capability.  Other than this, is there anything else we will need?  I think part of this will be making sure that all the data we need to show in the plots is put into the EVIO files and then translated into LCIO (even if this is only LCGenericObject collections).  Then the data will be available for plotting.  Otherwise, I don?t think we should be plotting anything that isn?t in the LCIO files, because this is likely to make the structure of the application a
>  mess.  (my opinion)
>> 
>> "shifter interface, monitoring app?
>> 
>> The process for a shifter will be loading the monitoring app, possibly with saved job and connection options, and then hitting the ?Connect? button.  Plots will start showing.  The shifter will watch the plots and will need to know how to interpret them.  That part can?t really be made any simpler.  Something more that would be needed there for a shifter interface?  I know you have mentioned direct system monitoring but I don?t think that belongs here.  This should be performed with sub-system tools for SVT, ECAL, etc. instead of the monitoring app, which is for looking at the final output events.
>> 
>> "Wired event display?
>> 
>> Can you clarify here?  What type of interface to Wired do you want?  Do we need a new JAS plugin to connect to the ET ring, pull EVIO events and translate them to LCIO as the monitoring app does?  Do you want to have different controls than the ?Go? buttons in the standard JAS GUI?  You can already use Wired offline with LCIO files, so I?m assuming this is for an online event display.  I need more detailed requirements here before I start to work on it.  Depending on the requirements, this could be a ton of work, actually.
>> 
>> "conditions system?
>> 
>> The classes and packages have been coded now for connecting to a database conditions system.  I think we now need to break this down further.  Omar and I discussed this yesterday.
>> 
>> The new conditions system with the database needs to be used in the SVT and ECAL codes.  I think these should be separate items.
>> 
>> We will need to setup a MySQL database at JLab that is externally accessible.  It will need to be loaded with conditions for the run.  Does this fall under the ?commissioning? item?
>> 
>> Thanks.
>> 
>> ?Jeremy

########################################################################
Use REPLY-ALL to reply to list

To unsubscribe from the HPS-SOFTWARE list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=HPS-SOFTWARE&A=1