lib/nsrep: refactor a piece of code

lib/nsrep: fix a bug in choosing when many timed out

Commit 270d9964 was wrong (released in >= 2.4.0). I had reviewed that change carefully, but still I didn't notice the problem with continue just above the changed hunk.

This is yet another logically equivalent change after 270d9964. Now it's written in a way that expresses the original intention more clearly and without copy&paste or long lines.

It seems easiest to verify by inspecting the meaning of the code separartely for the two cases, based on condition: cur_addr_score < KR_NS_TIMEOUT

I think this mistake contributed significantly to https://gitlab.labs.nic.cz/knot/knot-resolver/issues/470#note_105423

added bug label

@vcunat Any chance to write a test for this?

mentioned in issue #470 (closed)

I'm thinking of simple testing code that simulates the situation that a subset of IPs don't reply at all (configured probability + deterministic pseudo-random choice), by "forgetting" to send UDP packets to those IPs. That could serve as a simple basis for comparison of NS-choosing strategies.

@sbalazik @vcunat Please talk to each other :-)

added 14 commits

• 1546dace...fd15bc6c - 13 commits from branch master
• f13d3ef8 - lib/nsrep: refactor a piece of code

Compare with previous version

changed title from lib/nsrep: fix a bug in choosing when many timed out to lib/nsrep: refactor a piece of code

changed the description

assigned to @vcunat

Uh, I'm sorry for the confusion. I looked again, and all three code states were really performing exactly the same thing :-D now I refactored it again to yet another equivalent state that's supposedly "simplest".

Still, in the process I implemented that simple testing idea in eb90cd34, so we might utilize that work later.

merged

mentioned in commit fb8d0ca0