Hello Tim,
Do you have any idea if I use a lot of different question attempt step data will it have a performance impact ?
And when I say a lot, I really mean a lot, don't forget each formulas question can have an indefinite number of global variables, of local variables of subquestions, of answers per subquestion, of trials, ... !
For instance just my small tests from this morning trying to use separate data have created thousands of mdl_question_attempt_step_data records !!
Of course in Moodle 1.9 all this was serialized (no choice) so all the functions to serialize/ unserialize, recreate data structures from serialized strings are already done ! should I continue that way an only use one string per attempt ?
Should I separate each variable only serializing what is not a string (for instance the global variables instanciation must surely be serialized in a string like x0=1;y0=2;vx=4;vy=-4;tr=12;ta=18;)
Should I take a mid way using a fixed number of data serializing all similar values for all subquestions in one data for instance ?
I have absolutely no idea. All the question attempt step data is unknown territory for me. I don't even have a clear idea what names begining with -_ (or is it _- ?) are for ? I must clearly re-read what you have written in docs and the comments in the code !