Latest Results
Separate Artifact exchange and sharing in the Japanese Book (#1292)
## Purpose
Exchange a problem, an evaluated candidate, and analysis data without
requiring Experiment or an Adapter workflow.
## Changes
- Replace the solver-dependent Japanese Artifact tutorial with two
standalone scripts. The receiver starts in a new process with an empty
Local Registry and reads only the transferred `.ommx` file.
- Give layers, draft/committed objects, storage locations,
names/tags/digests, and the boundary between generic Artifact and
Experiment their own conceptual Guide.
- Separate authenticated registry sharing into a JSON-based how-to that
can be entered directly after SDK setup. Link the language-independent
Artifact specification for format details.
- Reconcile the affected learning contracts with the actual
introductions and direct-entry routes, separating basic Artifact
understanding from the file-roundtrip workflow. Keep concrete SDK
version guidance in setup.
- Extend the Book review skill to distinguish explanation gaps from API
friction, verify and deduplicate follow-up issues, and respect accepted
design tradeoffs without mixing SDK changes into Book PRs.
## Scope and stack
Part G of #1283, following merged #1291. Japanese Book only; no
SDK/API/runtime behavior changes. This PR targets `main` and is the
first active layer of [stack
#1297](https://github.com/Jij-Inc/ommx/stack/1297): #1292 → #1293 →
#1294 → #1295 → #1296.
The conceptual transformation Guides, Adapter tasks, Japanese
integration, and final English synchronization remain separate PRs. The
parent workstream remains incomplete until all slices and its
integration acceptance are finished. Teach recording and reopening experiments in the Japanese Book (#1291)
## Purpose
Record the conditions, inputs, and results of two small trials, save
them, and read them again without the Python objects that created them.
## Changes
- Replace the large Japanese Experiment tutorial with a six-cell
Notebook that evaluates two candidates, records the shared problem and
per-Run results, displays the recorded conditions, and reopens an
archive in a new process or kernel.
- Give Experiment/Run/Solve/Sampling relationships, explicit versus
automatic recording, status interpretation, and persistence their own
lifecycle Guide.
- Separate checkpoint restoration, fork, autosave, and standard
attachments into task-focused how-tos. Keep custom Codec and
backend-specific details out of first use.
- Align the reading-path ledger with the independent Experiment
workflow, removing unrelated Artifact, Adapter, and copy prerequisites
while retaining historical provenance and proposed-contract status.
- Update navigation and incoming Release Note links to the actual
content owners. Keep concrete SDK version guidance in setup
instructions.
## Scope and stack
Part F of #1283, following merged #1290. This PR now targets `main` and
is the first active layer of [stack
#1297](https://github.com/Jij-Inc/ommx/stack/1297): #1291 → #1292 →
#1293 → #1294 → #1295 → #1296.
Japanese Book only; no SDK/API/runtime behavior change. Artifact
exchange, conceptual transformation Guides, Adapter tasks, Japanese
integration, and final English synchronization remain separate PRs. This
PR does not close #1283. Latest Branches
0%
codex/constraint-results-by-kind 0%
codex/highs-dual-row-mapping 0%
codex/book-transformation-guides © 2026 CodSpeed Technology