Floating fees for alternative target spacings #5923

pull kroese wants to merge 1 commits into bitcoin:master from kroese:master changing 1 files +1 −1

kroese commented at 5:51 PM on March 18, 2015: none

The exponentially decaying window should depend on the configured target spacing

Floating fees for alternative target spacings

The exponentially decaying window should depend on the configured target spacing

42279e45a5

gmaxwell commented at 6:37 PM on March 18, 2015: contributor

I have no objection here, fewer magic numbers is good, but the behavior of the system may be completely ill-defined for arbitrary parameters. (Just a kind of silly example, a target spacing of 0 would otherwise work in the codebase AFAIK but here would result in everything turning to NaNs.)
kroese commented at 6:40 PM on March 18, 2015: none

Maybe I should change the line in: MAX(1, Params().TargetSpacing()) to prevent that?
morcos commented at 6:58 PM on March 18, 2015: member

I don't think it makes sense to change this. The command line argument that's passed is bytes per minute, and we're multiplying it by 10 a couple of lines below. You can't change the decay exponent here without fixing the comparison as well or changing the command line argument to be interpreted as "bytes per 1/10th of a target spacing".
kroese commented at 8:05 PM on March 18, 2015: none

@morcos Maybe you're right, but don't you agree it's strange to use a 10 minute window for this when the block duration could be much shorter (or longer)?
laanwj commented at 12:08 PM on March 19, 2015: member
- What exactly links this duration to the block duration, except that they both happen to be 10 minutes?
- This parameter is unrelated to floating fees as you state in the PR title. dFreeCount is used in the rate limiting of free transactions.
NACK
kroese closed this on Mar 19, 2015
DrahtBot locked this on Sep 8, 2021

Contributors