# Changelog

## Score

- CAI 52 → 60 (+7.6)
- Rubric changed (rubric-2026.10.5 → rubric-2026.11.2) — scores are not directly comparable.

## Lenses

- Code Health 64 → 64 (+0.1)
- Architecture 84 → 86 (+1.6)
- Maturity 34 → 49 (+14.8)
- Readiness 59 → 59 (+0.1)
- Security 79 → 91 (+12.0)
- Domain Modelling 100 → 100 (+0.0)
- Event Sourcing 100 → 100 (+0.0)

## Resolved (14)

- Ambiguous distinction between 'find' and 'load'. In many ORMs, 'find' implies a database lookup that might return nil, while 'load' might imply raising an error or loading from cache. Here, both take an ID and a class. Without documentation, it is unclear if they differ in caching behavior, error handling, or whether one loads projections/hydrates the aggregate differently. The naming suggests a semantic difference that is not immediately obvious from the signature.
- Ambiguous singular vs plural naming for similar retrieval operations. `load_event` takes a sequence number, implying it loads a specific version, while `load_events` takes only an aggregate ID, implying it loads all versions. However, the naming convention is inconsistent with `load_aggregate` (singular) vs `load_aggregates` (plural) in the Repository, and `load_events_for_aggregates` (plural). The distinction between 'loading a specific event by version' and 'loading all events for an aggregate' is obscured by the lack of a clear 'all' or 'history' keyword in the singular method name.
- Documentation: contradicts the code (docs/docs/concepts/aggregate-repository.md)
- Duplicate intent with different names. Both methods check for the existence of data associated with an aggregate ID. In event sourcing, 'stream' and 'events' are often synonymous in this context. Having two boolean check methods with different names for the same underlying concept is confusing.
- Inconsistent bang method usage. In Ruby, `clear!` often implies a more aggressive or irreversible action than `clear`, or simply that it bypasses some safety check. However, without clear documentation, it is ambiguous why two clear methods exist. If `clear!` is just a forceful clear, it should be documented. If they are functionally identical, it is redundant.
- Inconsistent naming for deletion operations. `permanently_delete_event_stream` deletes events. `permanently_delete_commands_without_events` deletes commands. The naming convention shifts from 'event_stream' to 'commands_without_events'. It is unclear if `permanently_delete_event_streams` (plural) is just the batch version of the singular, or if there is a semantic difference. The third method is very specific about what it deletes, while the first is generic about 'event stream'.
- No SBOM
- No artifact signing
- No build provenance
- Off-boarding risk: anonymized user #1
- Orphaned file (lib/sequent/generator/event.rb)
- Outdated: prism
- Outdated: web-console
- Unpinned build actions

## New (14)

- FileTooLong: migrations/projectors_replayer.rb (lib/sequent/migrations/projectors_replayer.rb)
- Off-boarding risk: anonymized user #1
- Orphaned file (integration-specs/rails-app/config/environments/development.rb)
- Orphaned file (integration-specs/rails-app/config/environments/production.rb)
- Orphaned file (lib/sequent/cli/app.rb)
- Orphaned file (lib/sequent/cli/sequent_8_migration.rb)
- Orphaned file (lib/sequent/core/helpers/association_validator.rb)
- Orphaned file (lib/sequent/core/helpers/attribute_support.rb)
- Orphaned file (lib/sequent/core/helpers/param_support.rb)
- Orphaned file (lib/sequent/core/helpers/secret.rb)
- Orphaned file (lib/sequent/core/helpers/string_to_value_parsers.rb)
- Orphaned file (lib/sequent/migrations/executor.rb)
- Orphaned file (lib/sequent/migrations/planner.rb)
- Orphaned file (lib/sequent/util/dry_run.rb)

## Changes since last survey

- 47 commits — 45 feature/other, 2 fixes

## By area

- lib/sequent — 19 commits
- (root) — 11 commits
- (repo) — 9 commits
- spec/lib — 5 commits
- docs/docs — 2 commits
- benchmarks/replay_persistor.rb — 1 commit

## Notable commits

- fix: Merge pull request #548 from zilverline/fix/ruby-regexp-cves
- fix: Move to the Ruby releases that fix the Onigmo regexp CVEs
- change: Add a benchmark for the replay optimized persistor
- change: Add before, after and around hooks to message handlers
- change: Allow specific write callbacks on projection records
- change: Allow the callbacks ActiveRecord adds itself through the default
- change: Always load the records using COPY
- change: Avoid allocations per record when tracking computed columns
- change: Check projection record callbacks in specs instead of in projectors
- change: Check that matching an InstanceOf returns a copy of its handlers
- change: Check that replaying flushes the own replay persistor of a projector
- change: Describe read-only projection records in the projector docs
- change: Describe the replayer changes for subclasses in the changelog
- change: Do not allow write callbacks or validations on projection records
- change: Do not let code other than projectors delete projection records
- change: Encode the COPY values with the C encoders of the pg gem
- change: Fail a spec when ActiveRecord adds a relation write method
- change: Inherit hooks that are added to a superclass later
- change: Invoke handlers directly without hooks, and skip after hooks on failure
- change: Let applications skip the projection record callback check
- …and 27 more

## Architecture

- Unchanged — 0 containers · 1 contexts · 0 edges
