THIS IS OBSOLETE, PLEASE SEE: https://twiki.cern.ch/twiki/bin/view/Atlas/PixCalWebBrowser

Introduction

This work is performed to provide and produce a "browsable"  summary of pixel calibration data. The tool was born out of the need to quickly access a set of information, common to a variety of scans, with a specified granularity.

...

NOTE: At the moment the DB access is slow and so it is not un-common for the request to take up to 20s so please be patient. We are working on speeding this up!

Task list

Tasks ready for training

• Open the browser in the ACR
  • Laptop on wifi?
  • Pixel shifters screen?
• Link from pixel wiki + link from shift instructions wiki (same?)

• INTIME_THRESH_SCAN exists twice in the drop-down menu that selects scan types.

• InLinkAnalysis
  • Does not show up in a date/time based search. Does show up with Scan# or Ana# search.
  • No bools of interest.
  • Variables I_pin_mod, I_pin_ped are listed in the table of summary of floats. They are not plotted. It would be nice if they were.
• BOC_THR_DEL_DSP (occupancy?) analysis
  • Too many bools. Remove ViSet_Low and ViSet_Ok(is that possible)
• Search by variable values continues to be desired.

General tasks

This section contains tasks already foreseen to go into future updates. The idea is that these tasks are mainly provided form the user community.

...

• Implement dynamic range of histograms based on entries filled. Maybe look at Util.cpp in the slac svn area?
• Implement extensive summary table always shown (at the top?)
  • Links to modules failing analyses (all, single, none of the analysis separately) PARTLY DONE
  • Link to modules that failed analysis
  • Make this list clickable so that one can directly look at the history of that module(future)
  • Show how the modules in the result results is distributed in the detector
    • Do this both for one analysis at a time in addition to all analysis together
    • Implement a detector module map that is read in. Basically a list of all modules in the detector, from excel file? We can then compare modules from the scans with this list Are there any SR1 vs PIT problems?
    • Fraction of detector scanned, fraction of B0/L1/L2/Disks scanned
    • Distribution in the detector of modules that failed?
  • Statistics for the given analysis: total and split up in analysis
  • • Maybe implement a standard 2D plot of module ETA vs module phi to display where the module are
• Display the result in a 2D plot of module ETA vs module phi
  • Example: ToT_mean for each module as a 2D histogram basedon the module eta/phi?Links to modules failing analyses (all, single, none?)
• Add timing statistics information PARTLY DONE
• Add quick choice buttons to list last months/weeks/days scans DONE
• Make sure that entering a scan and module name dumps info if compatible
• Move the module name input to the normal search to make sure the user understands that it can be used to select a set of scans/analyses DONE
• Put the calibration result and dump module info submit buttons below the list of analysis. They should only be visible when it makes sense to press them! DONE
• Implement search for analyses as well DONE** Change the input field to be of either scan or analysis DONE
  • Add radio button to select search for scan or analysis NOT DECIDED
  • Add drop down menu to look for analysis names, this should be made interactively so that selecting the radiobutton for analysis switches the type list. Interim solution is to have two type searches.
  • Show BOTH scan AND analysis date, add column
  • Make sure that they are sorted according to analysis (scan) if the search is for analysis (scan)
  • If the search is for analysis, table should list analysis number to the left and scan in later columns.
• If two analysis are looked at show an improved direct comparison summary
  • Link/list to modules that are different
  • Link/list to modules that are exclusive to either analysis
  • Make a statement on plot if the two are identical

• When selecting scans/analyses manually from the text box, mark all analysis that exist directly so that the user need only to deselect the analyses to be excluded
• Implement a variable fall-out menu for each calibration scan, understand why the javascript behaves so strangely for different browsers (Make sure that the makeplot can handle trend/overlay in this case)
• Add option to select modules/PP0(other object?) to compare in different scans (This should be merged with the detector browser navigation menu)
• Split all plots to show pixel types (were applicable from DB variables)
• Separate main calibration view with an expert output in the lower part
• Sort functionality using jqeury and CSS id's
• Clicking on analysis/scan link in list should submit results DONE
• With no query or selection the until date should be todays date (fix with javascript?)DONE
• Change the date format, skip the time, use only date DONE
• Implement a selection of scans/analysis by giving a scan or analysis ID DONE
• Finalize the detector browsable menu

Technical tasks

• The search for more than one analysis numbers do not work properly. It seems to select that last one in the list only. It seems to work when searching for scans though
• Add pointer to SimpleTiming class in order to allow for adding timing entries inside the class.
• Fix the setstate() function to handle all submits similar as there is no difference at the moment.
• Separate into single package to decouple from pixel DAQ DONE*
• Move code to CVS DONE
• Move to offline DB?
• Look into more efficiency file handling
• Clean-up and tag the plot files
• Look into the difference between the Firefox and Safari (+ other browsers)
• The plots are displayed in 3 columns but this is incinsistent if for one detector level there are no overlapping modules
• Only fetch the scan type selected from the DB
• Optimize the DB access to speed up the displays
  • Implemented new getAnalysis function in PixResultsCoral package which does 'OR' of connNames and and 'OR' of variables DONE
  • Try to make an OR of all the foreign keys obtained from the analysis ID table and check speed DONE
• Make sure that the FindMods function loops over only one variable!
• Optimize the browsable menu, at the moment it is slow due to multiple calls to DB
• Do I really need to copy around the result, why not keep the all_res as the member variable?DONE: NO GAIN IN SPEED
• Make a shell script that can supply the same input to the exe in order to quicker test the output without apache DONE
• Make sure all the scan names are transmitted when listing scans in dbutils (to remove the complicated current_scan_name business in the list) DONE
• Fix the unnecessary multiple query when asking for result from already specified variables, need to modify the sql query to remove multiple calls to the getAnalysisResultsFromDB DONE
• Remove the ul in the result section to save space DONE
• Fix bug so that when the 2nd or later analysis of the same scan is selected the correct name is output DONE
• Fix bug so that the scan type is correct even when not selected it in the drop-down menu or when it is summy (see above bug)DONE