Debugging errors that occur only in production code

There are a few techniques to debug errors that occur only in production:

1. Insert the following line:

MESSAGEBOX('<n>. ' + PROGRAM() + ' ...')

at "strategic" points in the program (e.g., before calling a function, method, procedure, etc.).  I start "n" at 1 and increment by 1 in subsequent MESSAGEBOX() calls.  The last number that displays in a MESSAGEBOX() before an error occurs helps me narrow down exactly where the problem is occurring.  If necessary, I delete the original MESSAGEBOX() calls and then insert new MESSAGEBOX() calls "closer" to where the error is occurring.  (Repeat as necessary [and, yes, you can -- and should -- try this at home <g>].)

2. Another technique is to write text to a log file (.Log, .Txt, etc.) at various points of/during program execution.  (This technique, by the way, is also useful for debugging COM objects since you cannot step through code in the VFP debugger nor can you [or are supposed to] send output to the screen from a COM object.)

3. In VFP 9, you can do coverage logging not only at design time but also in .EXEs (via the SET COVERAGE command).  Your client will need to send you the coverage log file  (generated by the SET COVERAGE command) which you can then run through the Coverage Profiler yourself.  Put code like the following toward the top of your main calling routine:

ON ERROR SET COVERAGE TO
SET COVERAGE TO <filename of your choice>.Txt