Local to me there are some jasmers for Massachusetts and New Hampshire called month-by-month challenges. These are basically Jasmer challenges except only for a smaller area. The checkers for these include the calibration for months not in available. As this could change over time, it seemed appropriate to find away around it. Examples below show this. Right now it only works for single Jasmer runs.
So this script went on and offline to fix some issues with how well it was scoring. It appears to be working now. I particularly like how it handles the Virtual and Swedish cache variants. The virtual shows how it handles discontinuous months reasonable well and it appears to find when caches started getting placed well.
Some samples are:
Virtual caches only jasmer
https://project-gc.com/Challenges//62459
Swedish caches only jasmer
https://project-gc.com/Challenges//62532
Performance wise it seems reasonable and does work worldwide. There might be benefit requesting more caches from a single call to GetOldestCaches with more caches to peel them off. It is not clear how this scales for when more than 1000 caches are published within the filter for a month. We might spend more time pulling useless data for recent months than we save for historic months.
It would be nice to add the ability to filter months by the total number of caches available for a month. This could avoid months where only one or two caches are available. Based on the current guidance for normal Jasmers, it appears 4 times more caches need to be available than are required. When implemented this could bring back double, triple, and greater Jasmers as months such as August of 2000 would be eliminated.
Feedback would be appreciated.