ccrook
commented
6 years ago snapspec station priority file will be CSV format file with columns:
code priority (integer - lowest numeric value wins) order (optional - maximum possible order of mark, overrides other mark order constraints from config file)
Add an optional file which assigns priorities to marks. These are just numbers that can be used to help select stations. The algorithm used by snapspec is described in:
https://htmlpreview.github.io/?https://raw.githubusercontent.com/linz/snap/master/src/help/help/programs/snapspec/algorithm.html
The updated procedure would apply at step 5. This will be modified so that the first criteria is the priority assigned to the station. Only stations with the highest numerical priority value (lowest priority) would be considered for failing at step 5. Stations with no assigned priority would be considered lower priority than all others.
The creation of the priority list can be as complex as we wish, to the point of a complete ordering of stations. However one likely approach would be to start with a criteria based on the first and last date of observation - more recently observed are preferred, and marks without older observations are preferred (as likely to be newer). We could also consider mark type (eg for benchmarks). The same file could also specify a highest possible order for each mark.
This would avoid adding a lot of complexity to snapspec, while giving us a lot of scope for defining intelligent criteria for assigning orders.