!editorial: simplify pseudo code (recovery) #3384
Conversation
Btw. shouldn't the weights be parameters and not fixed values (of 1/8 and 1/4)? And where does those values come from?
martinthomson
added
-recovery
editorial
ianswett
removed
the
editorial
| adjusted_rtt = latest_rtt | ||
| if (min_rtt + ack_delay < latest_rtt): | ||
| adjusted_rtt = latest_rtt - ack_delay | ||
| adjusted_rtt = max(latest_rtt - ack_delay, min_rtt) |
There was a problem hiding this comment.
This is the pseudocode that results from #3350, so this change is not editorial
There was a problem hiding this comment.
Right, this is a functional change.
Prior to this you get latest_rtt or latest_rtt - ack_delay.
With this change you get min_rtt or latest_rtt - ack_delay.
There was a problem hiding this comment.
Ups, yes, you are right. I think I got that impression from reading "MUST NOT apply the adjustment if the resulting RTT sample is smaller than the min_rtt", however, that's not what it says.
I think changing the if to the following would be slightly more readable for me but doesn't make a big difference:
if (latest_rtt - ack_delay > min_rtt):
Or even
adjusted_rtt = latest_rtt - ack_delay
if (adjusted_rtt > min_rtt):
adjusted_rtt = latest_rtt
However, we can also just withdraw this PR.
ianswett
changed the title
Btw. shouldn't the weights be parameters and not fixed values (of 1/8 and 1/4)? And where does those values come from?