Durga and I talked today about how to improve the current plots that we have and how to make them both more user-friendly for shifters and how to make sure people have a better understanding of what they need to look for/at.

Items from the discussion:

• Durga tried to run the online plots to remind himself what we've got and it failed due to a change in the data structure (Durga - feel free to correct me if I'm misunderstanding). Given this - the first modification will be to change things so that we can still get the plots we had in May.
  • Gene - you're going to have to do the same thing for the CKOV plots - so talk with Durga about that.
  • Chris - is the data structure finalized now? Or do we need to worry that this kind of thing will keep happening?

• We need better names for plots - less cryptic - something that actually explains what it is. This is true for both TOF and CKOV plots.

• Need better titles/labels/axis labels - again, more descriptive

• Need documentation for each plot that explains what the shifter is looking at and why. This is needed for both TOF plots and CKOV plots

• We want a better display - currently it's only possible to look at one plot at a time in a browser tab - ideally we want an array of important plots. We should have the default most important plots that are displayed together and then the ability to display others if necessary.
  • Chris - who is the right person to work on this? Would that be Alex?

We've currently got a deadline of September 15 for the first iteration of the improvements - this would allow it to be ready for the activation run and/or an Online/DAQ/C&M testing run prior to the October CM.

• Durga tried to run the online plots to remind himself what we've got and it failed due to a change in the data structure (Durga - feel free to correct me if
  I'm misunderstanding). Given this - the first modification will be to change things so that we can still get the plots we had in May.

Doh. Yeah, better put in some tests to make sure that the reducer runs and produces a plot...

• We need better names for plots - less cryptic - something that actually explains what it is. This is true for both TOF and CKOV plots.

Fine, but I strongly dislike the way we have this nasty index page and click on a link to view the png. I think that the plot name should not be the main identifier.

• Need better titles/labels/axis labels - again, more descriptive

Agree.

• Need documentation for each plot that explains what the shifter is looking at and why. This is needed for both TOF plots and CKOV plots

Agree.

• We want a better display - currently it's only possible to look at one plot at a time in a browser tab - ideally we want an array of important plots. We should have the default most important plots that are displayed together and then the ability to display others if necessary.
  • Chris - who is the right person to work on this? Would that be Alex?

Alex is the person to work on this.

What I was thinking

• We have a "main" page one for each detector containing the main plots. Should be visible on one screen.
• We have a default page that cycles through each "main" page, say, cycling between detectors every time "Next" link is clicked (or on a timer, say every 10 seconds?).
• We have a click through to access any more detailed stuff that isn't available on the "main" page for the detector. A bit more experty, should still attempt documentation etc nonetheless.
• We have an index of plots somewhere.

How do we handle reference plots? Do we need handling for reference plots for different reference run settings (e.g. with and without DS?). How do we manage that?

We've currently got a deadline of September 15 for the first iteration of the improvements - this would allow it to be ready for the activation run and/or an Online/DAQ/C&M testing run prior to the October CM.

Tight - I think we need to just get everything working again for that run. We can test all of this stuff offline.

The other thing I would add is a couple of "system" reducers - one to collect information on errors (is the reconstruction throwing up every spill for some reason?) and one to collect information on load - what is the CPU load, what is the memory footprint (and is this increasing i.e. do we have a memory leak anywhere?). Also add a display for log output.