< bitcoin-git>
[bitcoin] decryp2kanon opened pull request #20041: test: unit case sensitive in a functional test (master...test_btc_kb_200930) https://github.com/bitcoin/bitcoin/pull/20041
< sipa>
aj: you make a pretty compelling case to remove the "first" marker
< aj>
sipa: yeah :( it's sad, because it's cool
< aj>
sipa: hey, do you know how much disk it takes to compile the fuzzers? i ran out when i tried, with something like 20GB free
< bitcoin-git>
bitcoin/master 712f95d Aaron Clauson: Update msvc build to use new vcpkg manifest
< bitcoin-git>
bitcoin/master 36f5a58 MarcoFalke: Merge #19960: build: The vcpkg tool has introduced a proper way to use man...
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #19960: build: The vcpkg tool has introduced a proper way to use manifests (master...vcpkg_manifest) https://github.com/bitcoin/bitcoin/pull/19960
< bitcoin-git>
[bitcoin] decryp2kanon closed pull request #20041: test: consistently use BTC/kB and sat/B in wallet_basic test (master...test_btc_kb_200930) https://github.com/bitcoin/bitcoin/pull/20041
< bitcoin-git>
[bitcoin] sipa opened pull request #20044: Make all of net_processing (and some of net) use std::chrono types (master...202009_chrono_types) https://github.com/bitcoin/bitcoin/pull/20044
< bitcoin-git>
[bitcoin] kallewoof opened pull request #20045: upstream: add and use defaulted getters to UniValue (master...202009-defaulted-get) https://github.com/bitcoin/bitcoin/pull/20045
< bitcoin-git>
[bitcoin] laanwj closed pull request #19485: torcontrol: Create also a V3 ed25519-V3 onion address. (master...add-ed25519) https://github.com/bitcoin/bitcoin/pull/19485
< jnewbery>
Several travis builds seem to be failing with the last line ".......Running Unit Tests for Test Framework Modules" on the Arm64 builds. Anyone else seeing that?
< wumpus>
it looks like a timeout, "Ran for 1 hr 2 min 32 sec", not sure if that phase of the tests is specifically slow or it's simply Travis overloading their servers again
< vasild>
If we receive 10.0.0.1 from another peer we would not relay it but (if we have IPv4 connectivity) we would add it to addrman.
< vasild>
However, who would send us 10.0.0.1? If other nodes run the same code nobody would relay 10.0.0.1 to us, hmm...
< luke-jr>
vasild: no reason to assume other nodes run the same code
< bitcoin-git>
[bitcoin] theStack opened pull request #20047: test: use wait_for_{block,header} helpers in p2p_fingerprint.py (master...20200930-test-use-wait-for-block-header-in-fingerprint-py) https://github.com/bitcoin/bitcoin/pull/20047
< hebasto>
is an std::set a more proper container then std::vector for -bind arguments?
< hebasto>
^ wumpus vasild
< hebasto>
*than
< wumpus>
I guess if you want to detect duplicates
< jonatack>
Is anyone else seeing "Using default signet network" and " Signet derived magic (message start): ..." in the debug log, when launching bitcoind on mainnet/testnet/regtest?
< jonatack>
e.g. on master or when testing PRs based on post-signet master
< wumpus>
I haven't noticed it!
< wumpus>
jonatack: you're right!
< wumpus>
I see it too in my mainnet debug.log file now, the first two lines on every invocation
< wumpus>
it's not a terribly serious issue but it would be nice not to have in the release :)
< jonatack>
thanks! will push a fix
< jonatack>
looks like util_tests/util_GetChainName need updating too
< bitcoin-git>
[bitcoin] jonatack opened pull request #20048: chainparams: do not log signet startup messages for other chains (master...signet-startup-logging-bugfix) https://github.com/bitcoin/bitcoin/pull/20048
< vasild>
luke-jr: yes, but still a bit strange
< vasild>
hebasto: is it the case that the operations we do on the container are: insert (more than once), iterate over all elements? Then either one fine from complexity pov but wumpus is right that std::set would avoid duplicates.
< vasild>
no wait, I mean unordered_set
< hebasto>
do we need duplicated -bind ?
< vasild>
no
< hebasto>
for unordered_set we lack hash function
< vasild>
insertion in std::set is O(log(n)), in std::vector and std::unordered_set is O(1)
< vasild>
then KISS - use vector and (I guess) emit an error like "bind: address already in use" if the user supplies a duplicate
< hebasto>
ok, this set is too small, and using std::set makes code simpler
< vasild>
I guess this is the current behavior
< vasild>
you are right, we are going to put just a bunch of entries, complexity is irrelevant
< luke-jr>
if we don't need to do lookups, why do we need a set?
< phantomcircuit>
sipa, iirc there's some hard limit, possibly he's just got dbcache so large that it's exceeding the limit
< sipa>
phantomcircuit: that's possible, but the fact that he no longer observes an "is shutting down" message makes me wonder if there is more going on