When fAllowFree is True, nMinFee is incorrectly calculated #2874

Should help to stop "correct horse" type tx.

as @gmaxwell say, it may be not a bug. I'll cancel the request if more core devs reject me...

Please don't describe fee policy changes as "fixing a bug".

I don't think we should make this change. Further penalizing >0.01 BTC outputs seems unnecessary in light of the fact that very tiny outputs are not allowed, and really large dust paying transactions will not qualify as free.

@gmaxwell or only apply this check for incoming (mode == GMF_RELAY) tx? I think for relaying, (1+nsize/1k) * nMinRelayTxFee should be required.

Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/6f5b5be7f8adab9ee49f5e93542d2be595fea564 for binaries and test log. This test script verifies pulls every time they are updated. It, however, dies sometimes and fails to test properly. If you are waiting on a test, please check timestamps to verify that the test.log is moving at http://jenkins.bluematt.me/pull-tester/current/ Contact BlueMatt on freenode if something looks broken.

the size per KB is required for relaying, except for transactions which are small enough and have high enough priority to qualify as free.

This code and change are pretty hard to follow, but I think the end result here isn't especially unreasonable.

At a minimum this needs a lot more documentation, either in commit message or code comments or both. No ACK without that.

@jgarzik @luke-jr @gmaxwell I no longer expect this to be merged. It may be still useful for pool operators as a new rule for tx relaying and should help reduce the mempool size, and not affecting most of legitimate normal tx that is smaller than 1k bytes. Thanks for reviewing.