<bitcoin-git>
bitcoin/master 27f1121 fanquake: ci: drop -priority-level from bench in win cross CI
<bitcoin-git>
bitcoin/master 4a96486 merge-script: Merge bitcoin/bitcoin#32288: ci: drop -priority-level from bench in win cr...
<bitcoin-git>
[bitcoin] fanquake merged pull request #32288: ci: drop -priority-level from bench in win cross CI (master...win_cross_ci_bench_no_prio) https://github.com/bitcoin/bitcoin/pull/32288
<sipa>
Suhas is running benchmarks on historical mempool replay data to see if the spanning-forest algorithm could reasonably be a drop-in replacement for the current code
luke-jr has quit [Ping timeout: 252 seconds]
<sipa>
It seems very fast on realistic clusters (20-30 us for linearizing optimally anything up to 64 transactions, with some variance), though the worst case is likely a lot worse, but it can probably run just until time runs out and get something acceptable.
<instagibbs>
cool!
<sipa>
The min-cut algorithm (from the 1989) has a cubic worst-case complexity, but seems somewhat slower than spanning forest in practice (2-3x), and extrapolating suggests its worst-case runtime is in the 5-10ms range, which is too slow to run at relay time anyway.
pablomartin has quit [Ping timeout: 276 seconds]
<sipa>
+paper
<glozow>
very cool
<sipa>
Also, it's less practical as it can't be bootstrapped from an existing linearization to improve, and it's harder to just stop arbitrarily and get something reasonable out.
<sipa>
(which spanning-forest can)
<sipa>
That's it for me.
<achow101>
#topic Erlay WG Update (sr_gi, gleb, marcofleon)
<sr_gi[m]>
I finished re-writing the functional tests for the full-erlay implementation and CI is green both for it (#30277) and also the current PR to review (#30116). Some additional functional tests may come later, but for now at least we can make sure the Erlay flow is working as expected.
<sr_gi[m]>
I'm currently setting up Warnet to confirm the simulated results with real nodes. I’ll be reporting on the findings in the upcoming weeks. For people interested, you can take a look at the PR to review (#30116), which is taking care of selecting what peers to fanout or reconcile transactions with, and storing them in the right sets (no networking). You can also compile a full-erlay enabled node with the code in (#30277)) if you’re
<sr_gi[m]>
curious or you’’d like to try running that in the wild, though that is missing more extensive review.
<achow101>
#topic orphan resolution WG Update (glozow)
Emc99 has joined #bitcoin-core-dev
<glozow>
I'm just about done with TxOrphanage rewrite as multi index container + new eviction algorithm. I think it'll actually be more intuitive for reviewers than the current implementation, and my plan is to introduce txorphanage_impl.h, hook it up to all the existing txorphanage tests, and then change TxOrphanage to be a wrapper for TxOrphanageImpl. Still have more tests I want to write though. Hopefully something up soon.
<glozow>
Also have a list of potential extra things and I'm not sure if they're worth doing immediately
<glozow>
That's all
<sipa>
So with that you'd be able to just replace the TxOrphanage implementation, and the tests keep working?
<glozow>
Yeah pretty much, with a few modifications
<glozow>
For example, current situation has some randomization, but it'll become deterministic with the new algo
<sipa>
Right.
<glozow>
I think this could be easier to review than requiring people to learn the old TxOrphanage (which is kind of convoluted) and then compare it with the new one
TheRec_ has joined #bitcoin-core-dev
TheRec has quit [Read error: Connection reset by peer]
<stickies-v>
very excited to see the rewrite, agreed that the old one was pretty tricky to review
<achow101>
Any other topics to discuss?
<glozow>
I figured out how to templatize boost unit tests, so I can run the same test on <TxOrphanage> and <TxOrphanageImpl>, it's kinda cool
<glozow>
v29.0 is out!
<achow101>
\o/
<glozow>
must be time for 28.2
<achow101>
#endmeeting
<corebot>
achow101: Meeting ended at 2025-04-17T16:19+0000
<johnny9dev>
for gui-qml: back from 1.5 weeks of travel. Will be focused on answering questions around shutdown that came up and getting bitcoin-core/gui-qml#448 merged.
<bitcoin-git>
bitcoin/master 2835216 Pieter Wuille: txgraph: make GroupClusters use partition numbers directly (optimization)
<bitcoin-git>
bitcoin/master 247e9de merge-script: Merge bitcoin/bitcoin#32191: Make TxGraph fuzz tests more deterministic
<bitcoin-git>
[bitcoin] glozow merged pull request #32191: Make TxGraph fuzz tests more deterministic (master...202504_txgraph_deterministic) https://github.com/bitcoin/bitcoin/pull/32191
<bitcoin-git>
[bitcoin] sipa opened pull request #32300: feefrac: avoid integer overflow in temporary (master...202504_feefrac_div_fix) https://github.com/bitcoin/bitcoin/pull/32300
Christoph_ has joined #bitcoin-core-dev
Christoph_ has quit [Ping timeout: 252 seconds]
Christoph_ has joined #bitcoin-core-dev
kevkevin_ has quit [Remote host closed the connection]
Christoph_ has quit [Ping timeout: 248 seconds]
Cory97 has quit [Quit: Client closed]
Cory97 has joined #bitcoin-core-dev
eugenesiegel has quit [Ping timeout: 240 seconds]
jespada has quit [Ping timeout: 248 seconds]
pablomartin has joined #bitcoin-core-dev
jespada has joined #bitcoin-core-dev
pablomartin has quit [Remote host closed the connection]
Christoph_ has joined #bitcoin-core-dev
Guest60 has joined #bitcoin-core-dev
Guest60 has quit [Client Quit]
Christoph_ has quit [Ping timeout: 248 seconds]
Guest60 has joined #bitcoin-core-dev
Guest60 has quit [Client Quit]
<bitcoin-git>
[bitcoin] l0rinc closed pull request #31923: refactor: use original log string when no suspicious chars found (master...lorinc/optimistic-log-escape-message) https://github.com/bitcoin/bitcoin/pull/31923
<bitcoin-git>
[bitcoin] instagibbs opened pull request #32301: test: cover invalid codesep positions for signature in taproot (master...2025-04-codesep_cov) https://github.com/bitcoin/bitcoin/pull/32301
robszarka has quit [Quit: Leaving]
szarka has joined #bitcoin-core-dev
Robotico has joined #bitcoin-core-dev
Robotico has quit [Client Quit]
jespada has quit [Ping timeout: 260 seconds]
abubakarsadiq has quit [Quit: Connection closed for inactivity]