mcrossley wrote:having "null" values in the data file - In mine I have set all those to zero values
Maybe, it is just personal preference, as I'm not a computer specialist, but "garbage in, garbage out" is a saying that I have heard. I don't want to apply validation in each retrieval script, I prefer to apply it once when loading the database (and therefore I use nulls extensively). The use of zero values does not make sense to me, zero is definitely a valid number in many measurements.
Back in 2014 when I made my first exploratory venture into PHP and database, I rejected using your script (earlier version) and wrote my own because I wanted to use official nulls for any values where either the measurement was missing or it was suspect. Ironically it was some rogue temperature measurements that influenced my thinking, so using zero would definitely not suit me. Incidentally, in my script I also set time-stamps to null when the associated value does not "make sense" (unlike Cumulus that would leave those time-stamps at rollover time). So if there is no rain in a day, the time of the highest hourly rain (or time of peak rain rate) is to my way of thinking null rather than rollover time just as much as it would be if the rain gauge was not working. Of course, my database tables are not exact mirrors of any Cumulus log files, so I don't think your binding approach would work with my tables, and because I access real-time log directly on the rare times I need it, speed of loading database tables does not matter to me.
)