SRV ordering #32
Description
It seems that the ordering of SRV doesn't exactly follow the specified RFC 2782, see "Weight" on page 3.
Line 210 in 3fc6dad
The RFC suggests ordering by weight and then picking a random entry based on the incremental sum of weights. I don´t think this is equivalent to what is implemented here, but I am not sure how much the result differs.
Potentially more critical is that the current implementation seems to make Less(...) non-deterministic, this may have weird consequences, depending on the sorting algorithm that is used.