< bitcoin-git>
bitcoin/master fa02b47 MarcoFalke: refactor: Use AbortError in FatalError
< bitcoin-git>
bitcoin/master 39bd9dd MarcoFalke: Merge #19295: refactor: Use AbortError in FatalError
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #19295: refactor: Use AbortError in FatalError (master...2006-refactorAbortError) https://github.com/bitcoin/bitcoin/pull/19295
< bitcoin-git>
[bitcoin] fanquake opened pull request #19301: build: don't warn when doxygen isn't found (master...no_warn_doxygen_missing) https://github.com/bitcoin/bitcoin/pull/19301
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #19302: test: Check that peer is disconnected for bad buffered message (master...2006-badBuffer) https://github.com/bitcoin/bitcoin/pull/19302
< jonatack>
Following the addition of mypy to the functional test dependencies in bd7e530 and 5d77549, do the various build docs or others need to be updated?
< jonatack>
until I installed it, running the linters was showing "./test/lint/lint-python.sh: line 105: mypy: command not found"
< jonatack>
when running ./test/lint/lint-python.sh
< jonatack>
fanquake: yes, that is the second commit i mentioned. Wondering if the build docs or test docs should include mypy.
< fanquake>
isn't that in the test docs
< fanquake>
I'm not sure the build docs need to mention an optional python linter requirement.
< jonatack>
fanquake: thanks
< achow101>
Anyone else seeing "/usr/bin/ld: bitcoin_wallet-bitcoin-wallet.o:(.data.rel.ro+0x8): undefined reference to `InitError(bilingual_str const&)'" on master?
< bitcoin-git>
[bitcoin] troygiorshev opened pull request #19304: test: Check that message sends successfully when header is split across two buffers (master...2020-06-test-partial) https://github.com/bitcoin/bitcoin/pull/19304
< fanquake>
achow101: I'm not seeing any issues at the moment.
< achow101>
hmmm
< achow101>
if I revert the latest merge, it works
< achow101>
and i've tried both make clean and git clean -fdx
< fanquake>
are you passing anything to configure?
< achow101>
only --enable-debug
< achow101>
ah, dropping --enable-debug "fixes" the problem
< provoostenator>
achow101: on macOS I can build a freshly cleaned master just fine: ./configure --enable-debug --with-incompatible-bdb --enable-werror
< provoostenator>
Maybe wipe ccache?
< provoostenator>
I'm a big "git clean -dfx" fan, it's faster and more thorough than "make clean", but you already tried that.
< bitcoin-git>
[bitcoin] MarcoFalke closed pull request #19302: test: Check that peer is disconnected for bad buffered message (master...2006-badBuffer) https://github.com/bitcoin/bitcoin/pull/19302
< jonatack>
achow101: yep, saw it even after make distclean, make clean
< jonatack>
provoostenator: agree on -dfx, though that wipes my bdb dir as well if it's in there
< achow101>
provoostenator: clearing ccache doesn't help
< provoostenator>
achow101: on Ubuntu 20.04 I get the same error
< provoostenator>
Don't we have at least one Travis instance running with --enable-debug?
< provoostenator>
Reverting #19295 helps for me as well
< bitcoin-git>
[bitcoin] achow101 opened pull request #19310: wallet: BerkeleyDatabase make BerkeleyDatabase::Create, CreateMock, and CreateDummy non-static functions (master...bdb-refactor-create) https://github.com/bitcoin/bitcoin/pull/19310
< elichai2>
Is there any site that stores an index to the logs here? i'm trying to search of a conversation a while ago (CC kanzure)
< jeremyrubin>
elichai2: I think there used to be but not sure :/ Maybe best bet is to download all and grep?
< elichai2>
Hmm aren't these individual txt files? Hmm so if no one knows of any I'll write a short script to download them all and then I'll grep over them
< achow101>
I think the descriptor wallets project (https://github.com/bitcoin/bitcoin/projects/12) can be closed. unless we want to add the sqlite stuff to it, but I think that would be better to go into it's own project
< achow101>
also hardware wallet stuff could be in a project?
< luke-jr>
#proposedmeetingtopic (for wallet meeting) revert #6550 (conceptually) - merkle branches stored in the wallet would be useful for pruned nodes
< luke-jr>
#proposedwalletmeetingtopic revert #6550 (conceptually) - merkle branches stored in the wallet would be useful for pruned nodes [w/ watch-only wallets]
< instagibbs>
luke-jr, would be nice even as an option, elements/liquid could use it
< luke-jr>
instagibbs: apparently it's an issue for Stratum/Electrum servers too
< sipa>
i think it would make sense to optionally enable that
< sipa>
also, storing the fee for wallet transactions may make sense
< luke-jr>
hmm, not sure
< sipa>
you may care about fee/rate of a transactions that you're not the only participant in
< luke-jr>
sipa: after they confirmed?
< sipa>
before
< luke-jr>
we should probably save [unconfirmed?] input txs too :/
< luke-jr>
and that even for normal wallets
< sipa>
hmm, possibly
< luke-jr>
after all we have a financial interest in seeing those confirm too
< phantomcircuit>
iirc the wallet used to think that it included the dependent transactions (just one level) but it didn't actually fill them so they were just default CTransctions
< luke-jr>
looks like we had previous txs until 0.9
< luke-jr>
and merkle branches until 0.19
< luke-jr>
phantomcircuit: o lol
< sipa>
yes, i vaguely remember
< sipa>
i think it should just store those dependent txn in the wallet, if at all
< sipa>
rather than as members in each of the in-wallet descendants
< sipa>
to avoid exponential blowup
< luke-jr>
maybe
< luke-jr>
refcounted perhaps
< luke-jr>
feels like CMerkleTx might have been the best way to do it too :/
< sipa>
why refcounted? in order to clean them up afterwards?
< luke-jr>
sipa: yes; I suppose unnecessary if we do switch to logdb or such
< sipa>
i don't see how that's related
< luke-jr>
but I don't see much progress on that front despite discussion
< luke-jr>
it's related because you can't clean up an append-only file ;)
< luke-jr>
and refcounting there would just increase the storage for each increment/decrement
< sipa>
logdb would occasionally sweep to delete overwritten entries
< sipa>
otherwise it'd grow unboundedly
< sipa>
but i have no interest in reviving that; i think sqlite is a much better option
< sipa>
if we want to clean up not-mine dependencies after they've confirmed we could just do an occasionally garbage collection too, regardless of storage format
< luke-jr>
sipa: unbounded growth is what I expected from logdb, but ok
< bitcoin-git>
[bitcoin] renepickhardt opened pull request #19314: refactor: Use uint16_t instead of unsigned short (master...akh_uint16_t) https://github.com/bitcoin/bitcoin/pull/19314
< phantomcircuit>
sipa, unless the wallet can make use of backend indexing i dont really see the point of using sqlite
< sipa>
phantomcircuit: bluntly, i think bdb is crappy unmaintained software that was never designed for our use cases - and we've worked around that by being stuck to a 10y old version and various hacks with flushing
< sipa>
maybe crappy is too strong a term; it's just not what we need
< sipa>
sqlite does - it does more than that too, but it's extremely well tested maintained software, with very strong compatibility guarantees
< sipa>
logdb... i don't think we have the intent of putting in the manpower required to maintain/qa/test across platforms and whatnot, even if it's a very simple thing
< phantomcircuit>
sipa, yeah agreed, the logdb stuff mostly ran into issues with abstractions not making it easy to decouple things
< sipa>
achow101 is interested in using some of the indexing features... i'm less convinced about that, but at least having that possibility is a nice to have bonus
< bitcoin-git>
[bitcoin] amitiuttarwar opened pull request #19315: [tests] Allow outbound & block-relay-only connections in functional tests. (master...2020-06-test-outbounds) https://github.com/bitcoin/bitcoin/pull/19315
< bitcoin-git>
[bitcoin] amitiuttarwar opened pull request #19316: [net] Cleanup logic around connection types (master...2020-06-conn-refactor) https://github.com/bitcoin/bitcoin/pull/19316