< luke-jr>
vasild: I guess I mean with that change, -bind=… would result in the Tor-is-classified-as-non-Tor bugs coming back
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #20245: test: Run script_assets_test even if built --with-libs=no (master...2010-testAssetTestlibconsensus) https://github.com/bitcoin/bitcoin/pull/20245
< AdulrunaRedviva>
hi
< AdulrunaRedviva>

< AdulrunaRedviva>
How do I select the installed wallets to trade? (sample getbalance)
< wumpus>
AdulrunaRedviva: it's unclear to me what you mean, I suppose you are using multiwallet but what are yo trying to do? to select a wallet for RPC getbalance in bitcoin-cli you can pass -rpcwallet=<walletname>
< luke-jr>
AdulrunaRedviva: #Bitcoin for user questions, but rephrase it
< luke-jr>
(also note getbalance is broken in Core for the last several releases, but maybe not in a way you care about)
< wumpus>
ossifrage: strange, anything in debug.log?
< AdulrunaRedviva>
each user needs to make transactions for their wallet. How do I specify which wallet it is for when sending commands via rpc?
< luke-jr>
AdulrunaRedviva: again, #bitcoin
< AdulrunaRedviva>
okey.
< wumpus>
ossifrage: what it crashes on is a sanity check that the previous block of the block to be validated matches the current head block, I have no idea how you could end up in a state where that's not the case at that point in the code
< wumpus>
it's too bad this is only reproducible after 1h15m
< wumpus>
I don't think many people use verifychain to verify the entire chain
< luke-jr>
hmm, it looks like the /wallet/walletname endpoint isn't actually documented? :o
< wumpus>
documented where?
< luke-jr>
doc/
< luke-jr>
JSON-RPC-interface.md might be a good place
< wumpus>
yes it would
< vasild>
luke-jr: but now we have -bind=addr=onion to resolve those
< luke-jr>
vasild: I'm talking about the case where that is'nt specified
< luke-jr>
I wonder if Tor has a way to ask it, if a given source port is from Tor :P
< luke-jr>
I suppose the ideal for this would be to have Tor send us sockets over the control socket
< luke-jr>
not sure if Windows supports sending sockets tho
< vasild>
in #20234 maybe we can/should issue a warning if -bind= is given without -bind=...=onion and -listenonion=1 - something like "please specify -bind=...=onion in order to distinguish incoming tor connections"
< bitcoin-git>
[bitcoin] practicalswift opened pull request #20247: fuzz: Check for addrv1 compatibility before using addrv1 serializer. Fuzz addrv2 address serialization. (master...fuzzers-netaddr-post-addrv2) https://github.com/bitcoin/bitcoin/pull/20247
< bitcoin-git>
[bitcoin] dgpv opened pull request #20248: Fix test: length of R check in test/key_tests.cpp:key_signature_tests (master...fix-test-key-signature-low-r) https://github.com/bitcoin/bitcoin/pull/20248
< roconnor>
For those interested, I got back the result of instrumenting the compilation of "all of nixpkgs" with GCC 9. There are a handful of new packages miscompiled, again about half of them are test cases.
< roconnor>
Strangely BitcoinABC's CNetAddr:IsRoutable() gets miscompiled, but not Bitcoin's. I have no explaination.
< luke-jr>
O.o
< wumpus>
roconnor: I think what everyone wants to know most, anything at all potentially security sensitive?
< wumpus>
("yes" or "no" is fine, I understand if you don't want to give details)
< roconnor>
wumpus: Not that I have been able to tell. I still need to sort through the context of everything.
< wumpus>
okay :)
< roconnor>
There is a miscompiled crypto library in dolphin-emu.
< roconnor>
So maybe don't do you encryption on a gaming emulator.
< luke-jr>
lol
< roconnor>
mariadb-galera has some miscompilation in it gcache code. ... not sure how serious that is.
< roconnor>
oh an there is miscompilation in wireshark's packet-adwin-config.
< roconnor>
I'm not familiar with what ADwin is.
< luke-jr>
Wireshark has a history of vulnerabilities, so hopefully everyone is careful already :x
< wumpus>
thanks for looking into it, very interesting
< queip>
roconnor: uhm, stable releases of gcc sometime miss-compile well-formed C++ programs?
< wumpus>
compiler bugs happen quite often, though usually in more obscure circumstances, the surprise here was that it's so low-level, memcmp is used pretty much everywhere
< ossifrage>
wumpus, the debug log was just printing the progress report, it had made it to 50%
< ossifrage>
I recently moved my block data several times, finally ending on a SSD and I wanted to do a sanity check that everything is happy
< ossifrage>
Which is why I did a verifychain over all blocks
< sipa>
ossifrage: you're better off doing a reindex-chainstate in that case
< sipa>
checklevel=4 means rolling back to the point you specify, and then reconstructing everything after
< sipa>
if you're going to do that all the way back to genesis, just starting from scratch and only doing the forward part is going to be faster
< sipa>
by 50% you mean it did 50% of the blocks?
< ossifrage>
50% is just what the command reported in the log, no clue what that translates to
< sipa>
oh, 50% means it just made it all the way back to genesis in the rewind phase
< sipa>
that makes sense... it probably hit an edge case trying to rewind the genesis block
< ossifrage>
gmaxwell thought the assert might have happened on the genesis block
< sipa>
miscompilation in all GCC releases since 7.something
< luke-jr>
jonatack (IIRC) is right, the explicit fee modes are a mess :/
< luke-jr>
sipa: thoughts on <major>.<softforks>.<bugfixes>?
< bitcoin-git>
[bitcoin] luke-jr opened pull request #20250: Bugfix: RPC/Wallet: Make BTC/kB and sat/B fee modes work sanely (master...rpcwallet_explicit_fixups) https://github.com/bitcoin/bitcoin/pull/20250
< luke-jr>
jonatack: ^
< sipa>
luke-jr: not sure i like using consensus rules in the versioning scheme
< sipa>
that may make sense for libconsensus if that ever has separate versioned releases
< bitcoin-git>
[bitcoin] luke-jr opened pull request #20251: Move major version to first version integer as specified by SemVer (master...semver2) https://github.com/bitcoin/bitcoin/pull/20251
< luke-jr>
sipa: the code has consensus rules, though. This way, it's not only less confusing, but extra-clear to users what they're getting
< luke-jr>
to be clear, it isn't the consensus rules we'd be versioning, but Core's compatibility with them