I was looking at the Weather Bureau’s average charts, trying to see how far below normal this week is. Noticed a quaint leftover of ancient mainframe programming practices akin to Y2K.
Feb 29 is missing this year but present in leap years, so the form has to mark N/A in some way. It incorrectly fills in the expected numbers for high and low temps, but fills in 9999 for precip and snow.
Incidentally, this is a situation where NULL in C or None in Python would be the proper value for the variable.
In Python as an array of vectors:
Averages[28] = (44.0, 28.0, 0.06, 0.20)
Averages[29] = (None, None, None, None)
= = = = =
Later after looking it up: 9999 for missing data seems to be an established convention in weather records, specifically the Global Historical Climate Network.
Polistra's Mill 