< luke-jr>
fanquake: it forces BDB support off, no?
< luke-jr>
fanquake: eg, make depends NO_BDB=1 but with bdb4.8 on the system install
< luke-jr>
previously I assume the system BDB would have been picked up and used
< fanquake>
If that was the case then I think that was a bug, or at least not well documented that that would be the behaviour.
< sipa>
i would have assumed that a depends build never uses any system library ar all
< fanquake>
If you're building from depends you'd only want to be using the depends libraries
< luke-jr>
ok, I wasn't sure
< bitcoin-git>
[bitcoin] fanquake closed pull request #21764: build: produce Windows GUI build with Cirrus CI (master...cirrus-winbuild) https://github.com/bitcoin/bitcoin/pull/21764
< kallewoof>
I don't feel uncomfortable talking to luke-jr about the BIP editor thing, but from what I understand, luke-jr believes that there is an outstanding NACK against me becoming a BIP editor from gmax, which I believe is the reason for the standstill. Personally I don't think one person's NACK should be such a deciding factor, especially since the reason for the NACK doesn't even exist anymore, but I'm not going to push the
< kallewoof>
issue. If people want me to become BIP editor to help out I will gladly do so. My offer stands as it has for some time now.
< wumpus>
kallewoof: gmaxwell retracted the NACK it should be okay now
< kallewoof>
wumpus: That's my impression as well.
< vasild>
if it happens that we do the InactivityCheck() >60s after the node has connected, this means it took that loop >60s to process the other nodes (using non-blocking IO)
< vasild>
if that happens then InactivityCheck() will ask us to disconnect because node.nLastRecv == 0 || node.nLastSend == 0, ie nothing has been received or sent to the node for 60s after it connected. But this seems wrong because we would not have attempted to send or receive anything.
< vasild>
thus I think InactivityCheck() on just accepted nodes is undesirable.
< promag>
vasild: add an early commit that changes that behavior first by moving the accept connections to the end? as it is seems like an unintentional change
< vasild>
yes, I am looking into doing that
< promag>
vasild: great, note that I'm noob on net/p2p code
< vasild>
me too
< promag>
^^
< sipa>
if you guys keep improving the p2p/net code i'll soon be a noob at it again too ;)
< promag>
sipa: I thought you just do bitcoin stuff