💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400576119)
(Followup) should check that the hex string is of the correct length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400582914)
(Followup) should check length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400588522)
This seems unrelated?

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400590151)
(Followup) should check length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400589011)
(Followup) should check length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400595687)
(Followup) should check length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400597309)
(Followup) should check length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400596494)
(Followup) should check length

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400599589)
`Txid hash` ?

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#issuecomment-1820940555)
@maflcko perhaps we should add another `RPCArg` type that's 32 characters `STR_HEX`?

💬 dergoegge commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400613097)
`Txid::begin` etc. return `std::byte*` which can't be used for arithmetic operations.

See https://en.cppreference.com/w/cpp/types/byte.

💬 dergoegge commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400615735)
Looks like `ParseHashO` has a length check internally?

💬 dergoegge commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400618039)
Not sure what you mean here? `entry.first` is a `uint256`.

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400623243)
I was confused where `Txid` comes to play here. But I see `key` is a `COutPoint`.

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400626737)
I think I copy-pasted this comment in the wrong file. Can be ignored

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400632982)
Ah, it calls `ParseHashV` which indeed checks that it's 64 characters. So then only the `rest.cpp` line above doesn't check the length (and it's terribly documented).

💬 Sjors commented on pull request "Use Txid in COutpoint":
(https://github.com/bitcoin/bitcoin/pull/28922#discussion_r1400635225)
Update: it does, via `ParseHashV` - deleting the other comments below

💬 toolsopen commented on issue "sendrawtransaction takes too long":
(https://github.com/bitcoin/bitcoin/issues/28745#issuecomment-1821007250)
> Thanks for the update. Wallet performance is something that's being continually worked on.
>
> Did you end up upgrading versions (too), or still running on V22?

I also updated the final version to 25.1

💬 toolsopen commented on issue "sendrawtransaction takes too long":
(https://github.com/bitcoin/bitcoin/issues/28745#issuecomment-1821012378)
> Assuming this is a bdb wallet (can be checked via `getwalletinfo()[format]`), I don't think it is worth to spend time on optimizing the code, given that it will be removed in the future.

I use it for utxo maintenance and transaction broadcasting, but an overly large wallet will definitely have a great impact on performance.

💬 maflcko commented on issue "sendrawtransaction takes too long":
(https://github.com/bitcoin/bitcoin/issues/28745#issuecomment-1821019293)
What does the `getwalletinfo` RPC return in the `format` field, for you?

👍 theStack approved a pull request: "coins: make sure PoolAllocator uses the correct alignment"
(https://github.com/bitcoin/bitcoin/pull/28913#pullrequestreview-1742181205)
Tested ACK d5b4c0b69e543de51bb37d602d488ee0949ba185

Using [the emulated 32-bit ARM machine using qemu and Debian bookworm for armhf](https://gist.github.com/theStack/6eaeadd3fa1e521ed038b1ed93c101d6),
I could both reproduce issue #28906 on master and verify the fix in this PR. Having 2GB of RAM available without swap memory, bitcoind failed due to OOM on block height 322923 on the master branch, as the coins flush never happened up to that height due to an underestimation of the cache size.
...