Daniel-Constantin Mierla writes:
- this is intended to control the behavior of the tree data and matching
mode. The default one is 0, match longest prefix and the associated data with the prefix is a string. This is complete implementation.
There is actually a second mode, 1, intended to have the data as two integers separated by colon, like 'weight:value' iirc. The matching of any part of the prefix that has data associated with will add 'value' into a list of avps ordered by 'weight'. Since it was not really tested and the implementation might not be full -- due to lack of time at that moment and plans to migrate to use xavp for proper internal representation -- it was left undocumented. I should review this mode soon.
daniel,
do you see any problems with third mode that is combination of the previous two, i.e., where data associated with prefix is a string and its weight is determined automatically based on the length of the match (depth of the node in the three)?
-- juha