< bitcoin-git>
bitcoin/master f1f433e practicalswift: Make it easier to reason about node eviction by removing unused NodeEvicti...
< bitcoin-git>
bitcoin/master d94777b MarcoFalke: Merge #20302: net: Make it easier to reason about node eviction by removin...
< bitcoin-git>
[bitcoin] MarcoFalke merged pull request #20302: net: Make it easier to reason about node eviction by removing unused NodeEvictionCandidate::addr (CAddress) (master...NodeEvictionCandidate-addr) https://github.com/bitcoin/bitcoin/pull/20302
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #20316: test: Fix wallet_multiwallet test issue on Windows (master...2011-testWalletWindows) https://github.com/bitcoin/bitcoin/pull/20316
< jonasschnelli>
sips: as for the v2 AEAD. The packet length encryption context would still need its own keystream. I guess we can compute that stream on demand (4 bytes per message) and couple the rekey with the payload keystream.
< jonasschnelli>
Meaning. When we hit the rekey Limit we rekey also the length-encryption key but don’t precompute the length-encryption keystream since we don’t know how many messages fit.
< jonasschnelli>
(Meant sipa not sips)
< sipa>
jonasschnelli: yes, 3 streams; one for auth key, one for length encryption, one for data encryption
< jonasschnelli>
sipa: why does the auth key requires its own keystream? In the current construct, we just use the main payload keystream with block counter 0 (then 1+ for the actual encryption).
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #20318: build: Ensure source tarball has leading directory name (master...2011-buildTar) https://github.com/bitcoin/bitcoin/pull/20318
< wumpus>
so I think removing the milestone there makes sense
< MarcoFalke>
wumpus: Agree
< jnewbery>
hi
< hebasto>
agree too
< aj>
hi
< luke-jr>
aj: where bot
< wumpus>
#20284 was discussed in the P2P meeting, *something* like it needs to go in to make sure that previous versions don't parse the peers.dat as garbage and insert gerbage addresses
< achow101>
luke-jr: your pr would have introduced garbage that we would have to keep around forever. it may be simpler, but it definitely is no the correct way
< jnewbery>
I don't think we should be adding controversial features last minute to accommodate knots, which appears to be the main motivation for 20205
< wumpus>
if there's principal issues with the idea of having a unique id, I don't think opening another PR will resolve that
< jonatack>
20120 is a bugfix with 3-4 acks
< sipa>
#20120
< luke-jr>
jnewbery: this isn't adding features, it's NOT removign existing feature
< aj>
ugh, apparently some upgrade broke poor lightningbot
< meshcollider>
wumpus: it has an ACK too
< meshcollider>
And I am very nearly finished reviewing it
< wumpus>
ah no it has two normal ACKs as well
< wumpus>
thanks github
< luke-jr>
descriptor wallets should never have been merged without unique ids, it's a bug that they're missing
< jonatack>
heh github the ack hacker
< wumpus>
"hidden items" are for games not SCM interfaces
< wumpus>
#20153 is a bugfix and has two ACKs
< gribble>
https://github.com/bitcoin/bitcoin/issues/20153 | wallet: do not import a descriptor with hardened derivations into a watch-only wallet by S3RK · Pull Request #20153 · bitcoin/bitcoin · GitHub
< MarcoFalke>
18836 doesn't fix a regression I think, so it is fine to merge or not to merge, depending on review
< wumpus>
then there's #18818, which is unnecessary in my opinion, there hasn't been much review otherwise
< jonatack>
20305 contains a bugfix for the send rpc that can be moved to its own pull, if needed. The rest is really easier to do pre-release. Afterward, it would have to be overhauled a bit to support both the overloading and the new param.
< MarcoFalke>
The autogen.sh part was controversial and is not a regression-bugfix, so I've removed the milestone
< wumpus>
regarding the RPC units it's good not to introduce a RPC inconsistency for a release, so I agree we need to do something there
< wumpus>
MarcoFalke: +1
< jonatack>
If we don't want to overload conf_target and estimate_mode in these 6 RPCs, better to not release them
< jonatack>
than to have to support them and then deprecate them
< wumpus>
yes
< wumpus>
that would be silly
< jonatack>
so that was the motivation, sorry for doing it so late, but the merge yesterday of the PR 11413 follow-ups motivated me to spike on it
< luke-jr>
MarcoFalke: it is a bugfix for a regression in 0.20, but whatever
< MarcoFalke>
luke-jr: Yes, so it is not a regression in 0.21
< wumpus>
you're always very quick to call things bugfixes
< luke-jr>
wumpus: when they actually are, yes
< sipa>
what is the bug here?
< wumpus>
in any case we've been over the entire list now--that concludes this topic, happy reviewing
< luke-jr>
sipa: building goes looking for .git outside of the source tree and uses whatever it finds
< sipa>
ah
< sipa>
ok, i'd call that a bug, but a pretty minor one
< wumpus>
luke-jr: if it just fixed that, and touched nothing else, I'd agree
< wumpus>
still, it's not critcal enough to hold up 0.21 for in any case
< luke-jr>
20205 is tho
< wumpus>
any other topics?
< wumpus>
#endmeeting
< jonasschnelli>
#action: fix meeting bot! :)
< wumpus>
yess
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #20324: wallet: Set DatabaseStatus::SUCCESSS in MakeSQLiteDatabase (master...2011-walletSqliteSuccess) https://github.com/bitcoin/bitcoin/pull/20324
< luke-jr>
lightningbot: ping
< lightningbot>
pong
< luke-jr>
aj: thanks
< aj>
i don't think it actually supports meetings again yet