amesgen

Just to mention a workaround: Adding `Cabal` somewhere to the build plan fixes the issue. Seems related to #1836, in particular https://github.com/input-output-hk/haskell.nix/issues/1836#issuecomment-1422269562.

I think this is no longer an issue.

Another (orthogonal) idea to tackle this issue is to introduce refined notion of revalidation, tentatively called "slot-only revalidation", that is significantly cheaper and could be used in most cases instead...

> * This makes it somewhat awkward to change the contents of the candidate fragment without also changing the contents of the node's selection, which is an important part of...

FTR: See the PR description of https://github.com/IntersectMBO/ouroboros-consensus/pull/1288#issue-2604967619 for a concrete work item.

Some pointers regarding the topic of performance improvements in this area: - `sop-core` PR to represent `NS`/`NP` more efficiently: https://github.com/well-typed/generics-sop/pull/129 We mostly use strict versions of `NS`/`NP` in `strict-sop-core`, so...

Also: We have a relatively small upper bound on the number of eras we can support ATM regarding serialization: #328

The trace messages already contain enough detail to deduce the reason why a node changes its chain. Concretely, whenever we change our selection, we trace the so called "`SelectView`" of...

In both the old and new tracing system, you should see this when you enable JSON output and use a severity of at least `Notice` for `ChainDB.AddBlockEvent.AddedToCurrentChain`/`ChainDB.AddBlockEvent.SwitchedToAFork`.

> BTW: needed to edit your post to change my GH handle to @mgmeier - the correct one Thanks for catching that! > 2. In the new system, you'd additionally...