💬 glozow commented on pull request "cluster mempool: optimized candidate search":
(https://github.com/bitcoin/bitcoin/pull/30286#discussion_r1760336690)
Are these comments in the format timestamp, tx count, fee, size (or similar)? Or are they actually something like iteration count? I added a print statement to get feerates of clusters but got something slightly different.

```
total feerate: 4818977 / 54585, 71 txns
total feerate: 15029636 / 99008, 81 txns
total feerate: 1288344 / 98145, 90 txns
total feerate: 15755611 / 77992, 87 txns
total feerate: 429736 / 8245, 35 txns
total feerate: 695118 / 60056, 60 txns
total feerate: 941605 /
...

💬 sdaftuar commented on pull request "cluster mempool: optimized candidate search":
(https://github.com/bitcoin/bitcoin/pull/30286#discussion_r1760345789)
The format was timestamp (from my simulation), tx count, min iterations when running Linearize 10 times with different rng seeds and then once more when running it woth LIMO using the optimal linearization as input, and then max iterations over the same set of 11 runs.

💬 andrewtoth commented on pull request "refactor: prohibit direct flags access in CCoinsCacheEntry and remove invalid tests":
(https://github.com/bitcoin/bitcoin/pull/30906#discussion_r1760349038)
Since we don't have pass a bitmap as flags anymore, does it make sense to get rid of the `char` type for flags in `coins_tests` and replace with a more descriptive enum?
```C++
enum Flags
{
Clean,
Dirty,
Fresh,
DirtyAndFresh
};
```

💬 l0rinc commented on pull request "refactor: prohibit direct flags access in CCoinsCacheEntry and remove invalid tests":
(https://github.com/bitcoin/bitcoin/pull/30906#discussion_r1760375960)
what's the difference between clean and NO_ENTRY? Does that difference still make sense after this change?

> get rid of the char type for flags in coins_tests and replace with a more descriptive enum

Since we don't have flags in the production code anymore I would prefer getting rid of them in the tests as well. Reintroducing an enum would kinda' defeat that purpose in my opinion.

To make sure the tests are still checking the same combinations, I have added a few scripted diffs to trans
...

💬 andrewtoth commented on pull request "refactor: prohibit direct flags access in CCoinsCacheEntry and remove invalid tests":
(https://github.com/bitcoin/bitcoin/pull/30906#discussion_r1760378255)
`Clean` means not dirty and not fresh. `NO_ENTRY` is a special value that is not a possible flag but signals to the tests that that entry does not exist. With proper types, that would correspond to a `std::nullopt`.
We don't have to name the enum `Flags`, we can name it `EntryState`. That way we don't have to validate the `flags` variable to see if it's one of the values we will accept. That is better served by using an enum and std::optional.

💬 andrewtoth commented on pull request "refactor: prohibit direct flags access in CCoinsCacheEntry and remove invalid tests":
(https://github.com/bitcoin/bitcoin/pull/30906#discussion_r1760380040)
I don't think replacing with true and false is the right approach. Using enums, tests would be like
```
CheckAccessCoin(VALUE1, SPENT , SPENT , DirtyAndFresh, DirtyAndFresh);
CheckSpendCoins(ABSENT, SPENT , ABSENT, Fresh , std::nullopt );
```

💬 l0rinc commented on pull request "refactor: prohibit direct flags access in CCoinsCacheEntry and remove invalid tests":
(https://github.com/bitcoin/bitcoin/pull/30906#discussion_r1760380094)
> is a special value that is not a possible flag but signals to the tests that that entry does not exist

> That is better served by using an enum and std::optional.

Thanks, I'll investigate tomorrow.
Please leave some comments about the overall direction as well, do you think this new test layout is more readable or less?

👍 theStack approved a pull request: "streams: cache file position within AutoFile"
(https://github.com/bitcoin/bitcoin/pull/30884#pullrequestreview-2305633593)
Code-review ACK a240e150e837b5a95ed19765a2e8b7c5b6013f35

Didn't run any benchmarks to compare with master.

💬 theStack commented on pull request "streams: cache file position within AutoFile":
(https://github.com/bitcoin/bitcoin/pull/30884#discussion_r1760381984)
Possible follow-up material: now that that the read operation above (std::fgetc) has been replaced by only fseek/ftell calls on the file object, I think this condition can't be true anymore and hence this else-branch could be removed.

📝 l0rinc converted_to_draft a pull request: "refactor: prohibit direct flags access in CCoinsCacheEntry and remove invalid tests"
(https://github.com/bitcoin/bitcoin/pull/30906)
Similarly to https://github.com/bitcoin/bitcoin/pull/30849, this cleanup is intended to de-risk https://github.com/bitcoin/bitcoin/pull/30673#discussion_r1739909068 by simplifying the coin cache public interface.

`CCoinsCacheEntry` provided general access to its internal flags state, even though, in reality, it could only be `clean`, `fresh`, `dirty`, or `fresh|dirty` (in the follow-up, we will remove `fresh` without `dirty`).

Once it was marked as `dirty`, we couldn’t set the state back t
...

💬 glozow commented on pull request "cluster mempool: optimized candidate search":
(https://github.com/bitcoin/bitcoin/pull/30286#discussion_r1760397936)
aha, that makes much more sense!

✅ glozow closed an issue: "Mempool + Validation Code Organization"
(https://github.com/bitcoin/bitcoin/issues/25584)

💬 glozow commented on issue "Mempool + Validation Code Organization":
(https://github.com/bitcoin/bitcoin/issues/25584#issuecomment-2351846451)
closing as (1) my main gripe with the fee estimator has been resolved, and (2) there isn't 1 correct actionable plan here. There are some general ideas of separation that we should incorporate into e.g. cluster mempool stuff, but I don't think there's any point in keeping this issue open.

💬 edilmedeiros commented on issue "contrib/signet/miner: grind will fail for high difficulty chain":
(https://github.com/bitcoin/bitcoin/issues/30102#issuecomment-2351881650)
Confirming this behavior is still happening after #28417. I'll rework #30130 on top of it.

```
2024-09-15 12:39:41 INFO Mined block at height 10078; next in -5784h44m40s (mine)
2024-09-15 12:39:58 INFO Mined block at height 10079; next in -5784h42m28s (mine)
2024-09-15 12:41:01 INFO Mined block at height 10080; next in -5784h41m0s (mine)
2024-09-15 12:41:43 INFO Mined block at height 10081; next in -5784h39m12s (mine)
Traceback (most recent call last):
File "/Users/jose.edil/2-develop
...

💬 BenWestgate commented on pull request "Drop -dbcache limit":
(https://github.com/bitcoin/bitcoin/pull/28358#discussion_r1760434344)
> I'm not sure what "available" means here?

From [man free](https://man7.org/linux/man-pages/man1/free.1.html):

> **available**
> Estimation of how much memory is available for starting new applications, without swapping. Unlike the data provided by the cache or free fields, this field takes into account page cache and also that not all reclaimable memory slabs will be reclaimed due to items being in use

So having "enough RAM" will work but having "enough available memory" is a more accurate
...

💬 rob-scheepens commented on issue "Porting bcc tools to libbpf":
(https://github.com/bitcoin/bitcoin/issues/30298#issuecomment-2352217075)
@willcl-ark sorry for the delay in response. I'd love to implement this, but I doubt I currently have the necessary development skills (I'm just a performance engineer). I'll give it some thought and experiment a bit in the near future, let's see how far I get.

@0xB10C I just saw your eBPF Summit session, thanks for the update on the state of things. Have you by any chance seen the session on bpftime (https://youtu.be/YZbCBaTTkeE?si=a7qOB-arjxXBuTJP)? That appears to be useful in order to red
...

💬 maflcko commented on pull request "ci: honor ci bypass prefix in test-each-commit":
(https://github.com/bitcoin/bitcoin/pull/30898#issuecomment-2352219690)
> > > when opening or updating a PR that only touches a markdown file, I'd use this
> >
> >
> > We might still want to run spell check and such things for documentation changes.
>
> Insofar as the spelling linter in the CI doesn’t raise

There are many other linters that do raise, and even many that only serve to lint docs. If they serve no purpose, they should be removed, instead of being (required to be) skipped.

Regardless, many doc updates are only checked at compile time (doxyg
...

💬 willcl-ark commented on pull request "test: re-bucket p2p_node_network_limited":
(https://github.com/bitcoin/bitcoin/pull/30879#issuecomment-2352254067)
> The re-bucketed test in the commit doesn't match the one in the commit/PR title (p2p_timeouts vs. p2p_node_network_limited), so I suppose one of the two has to be adapted.

Ugh, I cherry-picked the wrong tests from my working branch. The title is correct, tests in the commit here updated.

> I think the sorting isn't really based on the "correct bucket", but rather the inverse overall duration, the buckets are just a guideline/reminder. That is, as long as the longest running test is start
...

💬 maflcko commented on pull request "test: re-bucket p2p_node_network_limited":
(https://github.com/bitcoin/bitcoin/pull/30879#issuecomment-2352259405)
Thanks, the explanation is correct and the two links with data are useful to support this change.

review ACK 45663cd02cfb99c0bf91828b33a0611bb223497e

💬 maflcko commented on pull request "test: re-bucket p2p_node_network_limited":
(https://github.com/bitcoin/bitcoin/pull/30879#issuecomment-2352270431)
Nit: May be good to reword the pull request description to say that this better reflects the reality in CI runs, and may even speed them up. The mention of "CI timeouts" could be removed, because adding or removing 200 seconds from a run shouldn't be a fix (or cause) of a timeout. Any run that takes longer than half of the overall timeout value is a warning that should be investigated and fixed, because the CI timeout is really only there to catch cases where there is a true timeout (zombie proc
...