< kallewoof>
Sigh. Why would the wallet_avoidreuse.py get a 'method not found' in travis when it works locally? I don't even know where to begin debugging that.
< kallewoof>
Could be an OS thing I guess. *tries on linux*
< ken2812221_>
kallewoof: You should skip it if wallet disabled.
< kallewoof>
I think you're right on. It says $ export DEP_OPTS="NO_WALLET=1"
< kallewoof>
I'm confused why other tests are working though
< ken2812221_>
See #14180
< kallewoof>
OK I see it now. It's the skip_test_if_missing_module thingie that is in other tests but not in mine. Adding it. Thanks for the hint!
< Jmabsd>
where again is the code that validates the signature for a segwit input? (as in, generates the signtext for a segwit input, either to validate an input or to generate a transaction)
< wumpus>
hebasto: sure! I see what went wrong there, it was backported to the 0.16 branch, and thus ended up on the exclude list for 0.17 (to make sure PRs only end up in the earliest release's release notes, in this case 0.16.3), however there hasn't been release on the 0.16 branch since. Anyhow, I'll add it in manually.
< wumpus>
hebasto: added
< hebasto>
wumpus: thank you
< hebasto>
What is a proper way to start working on someone's PR labeled "Up for grabs"? Open a new one?
< gmaxwell>
Does "Downgrade warning" need a warning about the txindex change? can you not go back from 0.17 if you use txindex?
< jimpo>
gmaxwell: Hmm, I think you could go back but the txindex would be missing. So you'd have to reindex to get it back.
< gmaxwell>
so if you go back and set txindex=0 you'll be okay? that should probably get mentioned.
< kostiantyn>
Is it still true? It seems like the node starts downloading blocks once it sees the header which passes nMinimumChainWork.
< sipa>
kostiantyn: for any given block, we download the header first
< sipa>
and then the block data itself
< sipa>
but the header downloding and block downloading in general run in parallel; it's just easier to describe them as two sequential processes
< sipa>
(they've always run in parallel, since headers-first was implemented in 0.10)
< kostiantyn>
thanks a lot for the explanation!
< kostiantyn>
maybe it makes sense to update this line in the wiki " Once the node has all of the headers, from the genesis block up to the current tip of the blockchain (380,000 as of October 2015), only then does it begins downloading the full blocks."
< gmaxwell>
There is no need to poll on an empty watching set.
< Lightsword>
gmaxwell, so that poll() bug shouldn’t affect us?
< sipa>
i don't understand the issue if it's just in case of an empty set... you can easily workaround that by detecting an empty set being passed in, and if so, just sleepign
< sipa>
either it's more than just that, or it's a pretty boring bug :)
< Lightsword>
ok, just wanted to make sure that wouldn’t cause problems, by the way, what was happening in regards to switching to libevent for node p2p sockets, did that effort stall for some reason? I see #11227 but it hasn’t been updated in a while
< gmaxwell>
sipa: I wonder if on ryzen where you have 2x parallelism in shani, if it would makes sense to use multiple cores to compute hashtrees... or if it couldn't overcome overheads.
< phantomcircuit>
so why is SOCKET unsigned int when all the fd parameters to things are int?
< * phantomcircuit>
goes to ask 2011 wumpus
< phantomcircuit>
any idea?
< phantomcircuit>
oh i see
< phantomcircuit>
unsigned on windows
< conman>
is 0.17.0 not tagged no git for a reason?
< conman>
I can only find v0.17.0rc3
< conman>
tag, that is
< sipa>
yes, because 0.17.0 is not released yet :)
< conman>
ah very good
< conman>
I misunderstood, thanks
< sipa>
don't worry, it's very close
< conman>
thank you, the posts I'd read referencing 0.17.0 didn't say "will have"; it sounded like it was out