< bitcoin-git>
[bitcoin] yuntai closed pull request #13365: RPC/REST/ZMQ, Wallet: Set label with importprivkey only requested (master...master) https://github.com/bitcoin/bitcoin/pull/13365
< bitcoin-git>
[bitcoin] qmma70 opened pull request #13388: util: Implement boolean conversion and !operator for uint_* (master...uint_bool) https://github.com/bitcoin/bitcoin/pull/13388
< mryandao>
i actually asked about patch submission via mailing list in response to the takeover rumors.
< wumpus>
mryandao: eek I understand now
< bitcoin-git>
[bitcoin] n2yen opened pull request #13389: Utils and libraries: Fix #13371 - move umask operation earlier in AppInit() (master...13371) https://github.com/bitcoin/bitcoin/pull/13389
< wumpus>
mryandao: I don't think there's a hurry to get away from github, but I do think the microsoft takeover is the slow road to obsolence, just like sourceforge, until the service is finally put out of its misery (like happened with microsoft's other code hosting service, codeplex)
< wumpus>
aanyhow maybe we should bring up the host-our-own-gitlab-instance topic again at next meeting
< wumpus>
at the least I'm going to cancel my paid github membership as asoon as the announcement goes through, having to pay microsoft-tax with laptops is enough, not going to give them any more money...
< wumpus>
(while I didn't mind supporting a smaller company that simply provides a, generally well-maintained, service)
< bitcoin-git>
bitcoin/master f41d339 practicalswift: bench: Use non-throwing ParseDouble(...) instead of throwing boost::lexical_cast<double>(...)
< bitcoin-git>
bitcoin/master 2722a1f Wladimir J. van der Laan: Merge #13383: bench: Use non-throwing ParseDouble(...) instead of throwing boost::lexical_cast<double>(...)...
< bitcoin-git>
[bitcoin] laanwj closed pull request #13383: bench: Use non-throwing ParseDouble(...) instead of throwing boost::lexical_cast<double>(...) (master...remove-dependency-on-lexical_cast-which-is-boost-and-also-throws) https://github.com/bitcoin/bitcoin/pull/13383
< wumpus>
so it expects "Requested wallet does not exist or is not loaded" but it gets a CalledProcessError
< wumpus>
"subprocess.CalledProcessError: Command '/Users/joao/Projects/bitcoin/src/bitcoin-cli' returned non-zero exit status 18."
< wumpus>
code 18 is RPC_WALLET_NOT_FOUND
< wumpus>
so that looks correct, I'm confused
< wumpus>
promag: just tried to run that test separately here on master (ubuntu 16.04), no issues
< wumpus>
promag: not sure what is special about your environment that could trigger this
< promag>
wumpus: thanks
< promag>
wumpus: I too can't understand why
< wumpus>
you could try doing a completely new checkout, then if that passes, compare the difference
< wumpus>
if even a new build from a clean tree doesn't pass, it would be extremely confusing
< promag>
right, I hate on this happens
< promag>
*when
< wumpus>
it's bad, but one level of badness below issues that reproduce on travis but not locally
< promag>
wumpus: git clone, autogen, configure, make, test same error :|
< promag>
Python 3.6.5
< * wumpus>
eats his hat
< wumpus>
promag: what OS?
< promag>
wumpus: macos 10.13.3
< wumpus>
can't try that, but I see ubuntu 1804 has the same version of python, will try that
< promag>
ok, ty
< promag>
I'll be back in a bit
< bitcoin-git>
[bitcoin] ken2812221 opened pull request #13390: Tests: Ignore RemoteDisconnected and BadStatusLine when stopping node (master...stop_node) https://github.com/bitcoin/bitcoin/pull/13390
< bitcoin-git>
[bitcoin] practicalswift opened pull request #13391: Make tinyformat noexcept: Use default error handling (assert(0 && reason)) on incorrect format strings (master...non-throwing-tinyformat) https://github.com/bitcoin/bitcoin/pull/13391
< jnewbery>
promag: seems like the bitcoin-cli stderr is not matching the expected string. See test_node.py L370. I suggest you put a breakpoint in there and see what cli_stderr is set to
< promag>
jnewbery: will do
< wumpus>
promag: passed on ubuntu 18.04 with 3.6.5
< wumpus>
+python
< promag>
right
< wumpus>
so yes makes sense to try to debug this issue locally, it's unlikely we're going to be able to reproduce it
< bitcoin-git>
[bitcoin] practicalswift opened pull request #13392: util: Make strprintf noexcept. Improve error message on format string error. (master...strprintf-noexcept) https://github.com/bitcoin/bitcoin/pull/13392
< marcoagner>
promag: got curious at lunch and successfully passed the test you're having issue on my other laptop (macos 10.13.4 with python 3.6.5)... no help, it seems.
< wumpus>
promag: you don't happen to have a different version of bitcoind running that it, for some reason or bug, runs against?
< jonasschnelli>
sipa: Yes. Right. SSE4 was enabled before..
< roasbeef>
jonasschnelli: external version for how to decipher the thing in the first place, internal version tells wallet what scheme was used to derive the set of addrs
< jonasschnelli>
roasbeef: I see. Thanks!
< gmaxwell>
sipa: I can but it'll be a bit since that computer is in a box right now.
< gmaxwell>
echeveria: At least a few weeks ago I was watching it and almost never seeing hits out of my ordinary sized extrapool... but that might have just been because it was a bit too small.
< echeveria>
2018-06-04 18:15:44.495322 Successfully reconstructed block 00000000000000000037acbd8ccb51a8f09222800eab38f30c89bbe48ade3925 with 1 txn prefilled, 2075 txn from mempool (incl at least 13 from extra pool) and 16 txn requested
< echeveria>
2018-06-04 18:06:29.634709 Successfully reconstructed block 0000000000000000000e62f80def774b415e812da4473cdb49f3ba56ea0f1865 with 1 txn prefilled, 343 txn from mempool (incl at least 11 from extra pool) and 332 txn requested
< echeveria>
woof.
< gmaxwell>
oh man, yea thats pretty good
< echeveria>
yeah I'll collect some proper stats.
< echeveria>
a lot of the time I'm getting complete reconstructions which is nice. the node is running on a core i3 box that's mega underpowered, intentionally to see how it'll run.
< gmaxwell>
sipa: should call for someone to try out 13386 on shani supporting intel, ... they only have it on some atom cpus.
< sipa>
i know; Goldmont only
< gmaxwell>
not that its that important but it would be interesting to see numbers.
< echeveria>
I looked around a while ago and couldn't find anything easy to get my hands on with a goldmont chip
< echeveria>
works on my rizen 7 server though.
< gmaxwell>
I assume DO or some other hosting provider is using goldmonts in vpses.
< echeveria>
they're larger x86 cores for DO and ScaleWay.
< gmaxwell>
(since basically thats the target application for that chip, AFAICT)
< echeveria>
ScaleWay has armv7 servers which is vaguely interesting, I run an edge router node on one of their ARM boxes.
< luke-jr>
what is the cpuinfo entry for shani?
< luke-jr>
a: sha_ni
< sipa>
luke-jr: sha ?
< sipa>
sha_ni, indeed
< sipa>
on my Ryzen
< cfields>
sipa: taking an optim from the intel sha2 whitepaper gives me a ~6% speedup on the SHA256D64_1024 bench, using sse4 (no avx2 here :( )