💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469609534)
Why not? The InputFetcher should not make assumptions about the structure of the block being passed in.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469616204)
Why don't we want different benchmarks on different machines? All benchmarks are subtly different depending on the host machine.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469618569)
Fixed in latest iteration.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469625970)
0 threads means it is turned off, yes. If `-par=1` is configured, we want to pass 0 to input fetcher to disable prefetching. Do you have a suggestion of how this can be made more clear?

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469631434)
Possibly.
This is no longer relevant in the current implementation.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469638986)
No, we can have no worker threads, for instance if started with `-par=1`. In that case just disable prefetching.

💬 CMDRZOD commented on issue "sqlite legacy descriptor wallet migration fails":
(https://github.com/bitcoin/bitcoin/issues/33468#issuecomment-3456559038)
I am also experiencing this same issue with v30. I downloaded compiled binaries to use on ubuntu 22.04

I also tried upgrading a backup of the wallet using windows binaries on Win 11 which also failed with the message unable to rescan.

My data dir is in a custom location and am unable to get the bitcoin-wallet dump command to work. What is the recommended path forward?

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469655686)
When do we do a batch write though in this example?

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469658436)
Done.

💬 brunoerg commented on pull request "addrman, net: filter during address selection via AddrPolicy to avoid underfill":
(https://github.com/bitcoin/bitcoin/pull/33663#discussion_r2469690257)
Thanks for adding test coverage. By doing mutation testing for this PR, I noticed that the following mutant has not been killed (it means tests were not able to detect this change):

```diff
diff --git a/src/addrman.cpp b/src/addrman.cpp
index c06fbb0673..4cc7073c3b 100644
--- a/src/addrman.cpp
+++ b/src/addrman.cpp
@@ -854,7 +854,7 @@ std::vector<CAddress> AddrManImpl::GetAddr_(size_t max_addresses, size_t max_pct
// Filter by policy
if (policy && policy(ai)) {

...

💬 furszy commented on pull request "interfaces: enable cancelling running `waitNext` calls":
(https://github.com/bitcoin/bitcoin/pull/33676#issuecomment-3456619672)
Code review ACK https://github.com/bitcoin/bitcoin/commit/dcb56fd4cb59e6857c110dd87019459989dc1ec3

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469691257)
Done, using uint32_t in the Input struct now, not sure if it will have any effect in the struct though.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469694470)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469695146)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469695839)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469696287)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469697562)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469698872)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469699476)
Done.

💬 andrewtoth commented on pull request "validation: fetch block inputs on parallel threads >10% faster IBD":
(https://github.com/bitcoin/bitcoin/pull/31132#discussion_r2469710016)
> Does this mean the spent tx is never processed on the same thread currently?

I don't think that's what it means. The same thread could fetch two inputs in a row.

> Maybe we can mix it up a bit by something like

So we can use the same prevhash in difference txs? What is the benefit of this?