Potentially non-deterministic simulation problems caused by SQLite CCDB files??

[sdobbs]

@aaust pointed out that he was running into some problems with his simulations that were hard to track down.

A breakthrough came when the following error was noted, from a fresh CCDB SQLite file:

ifarm1802.jlab.org> ccdb log
(sqlite3.OperationalError) Could not decode to UTF-8 column 'logs_comment' with text '

 BCAL data base of constants used in MC

 Silicon PMT dark rate (from GlueX-doc-1754-v3)
 Single PE width to amplitude ratio
 6.489/61.67

The problem was traced down to assignment 71 (the only one to /BCAL/bcal_parms).
The comment was set to the empty string, and the (long) comment is captured below.
This solution is being checked.

The current main questions are: (1) when did this problem start; (2) why did this problem start; (3) what is even causing this problem?

Note that this does not seem to be a problem when working with MySQL.

 BCAL data base of constants used in MC

 Silicon PMT dark rate (from GlueX-doc-1754-v3)
 Single PE width to amplitude ratio
 6.489/61.67 = 0.105 (from GlueX-doc-1754-v3)
 Pedestal width to single PE amplitude ratio
 8.585/61.67 = 0.139 (from GlueX-doc-1754-v3)
 Variation of gains between individual SiPM tiles.
 Yi measured a value of ~4% and indicated this in
 private communication.
 This factor includes both the cross-talk and after-pulse
 probability for a 100ns window. The value from GlueX-doc-1754-v3
 was 15.7%, but is for a 1 us window. Private comminucation with
 Yi indicated a value of 12% from previous measurements with a
 200 ns window. Note that the 100ns assumed in the current simulation
 has the actual pulse starting 25ns into that making the actual
 cross-talk/after-pulse window at 75ns. Estimatation from Yi for
 a 100ns window is ~10%, hence the value used here.
 integration widnows in [ns]
 Photon detection efficiency (from GlueX-doc-1754-v3)
 sampling fraction:
 The value of 12.5% from NIM A 596 (2008) 327?337 (a.k.a. GlueX-doc-1071)
 is superceded by Stamatis' thesis value of ~9.5% due to the proper
 inclusion of the (inactive) cladding of the fiber
 Digitized values due to dark pulses only
 Inner region has 3 digitized values/event
 Outer region has 2 digitized values/event
GX-doc 1069, Table 1 -- try to extract back to
photons per side per MeV in fiber
4.6 / PDE / attentuation  (meaurements performed in center)
75 = 4.6  / 0.12 / exp( -200 / 300 )
set the sampling smearing coefficients:
(from GlueX-doc 827 v3 Figure 13 )
time smearing comes from beam test time difference resolution with
beam incident on the center of the module
no floor term, but leave the option here:
 Threshold applied to energy deposition in MeV (HDGeant)
 time window for two hit resolution in ns
 maximum number of hits per readout channels