HCRM · Computational reproduction

Merge changes.
Keep the context.

Run the original eight cases from WinMedia’s HCRM evaluation demo. See which edits survive, which conflicts stay visible, and which histories are rejected.

Pinned package v0.18 original casesNo personal input

Execute & compare

This runs the pinned Python package on the MandalaStacks server and compares its output with the archived reference.

No fresh run yet. Inspect any case below before starting.

No account, upload, or saved history. Download the receipt to retain this run after leaving the page.

Inspect the eight original cases

Select a case to compare its inputs and result.

Case 1 / 8

Disjoint concurrent edits survive

Alice changes the owner. Bob changes the status. Both edits should survive the merge.

Inspect the original input histories

[
  [
    {
      "op_id": "a1",
      "actor": "Alice",
      "seq": 1,
      "deps": [],
      "kind": "SET",
      "key": "owner",
      "value": "Alice"
    }
  ],
  [
    {
      "op_id": "b1",
      "actor": "Bob",
      "seq": 1,
      "deps": [],
      "kind": "SET",
      "key": "status",
      "value": "review"
    }
  ]
]

Archived reference

Expected merged state

The merged record keeps owner “Alice” and status “review”.

View archived output

{
  "case": "Disjoint concurrent edits survive",
  "result": "PASS",
  "state": {
    "values": {
      "owner": "Alice",
      "status": "review"
    },
    "counters": {},
    "conflicts": {},
    "op_ids": [
      "a1",
      "b1"
    ]
  },
  "checks": [
    "expected state",
    "reverse order",
    "duplicate delivery"
  ]
}

Fresh actual output

Run the experiment to produce a new result. The archived result is not a fresh execution.

Reproduce it yourself

Run the original bundle

Use the same pinned package independently, including when the website’s Python runtime is unavailable.

Download the original bundle Read the original README

Unzip the downloaded file.
Open a terminal in the extracted hcrm-demo-candidate folder.
With Python 3.9 or newer available, run:

python3 -I -S run_demo.py

Compare the resulting JSON with EXPECTED-RESULT.json. No extra Python packages are required. Expected invalid-input rejection is a passing result.

Verify the pinned package

Bundle SHA-256:

e60cca94df63460d35f68c522a0ec61e0b8b0baf2f0fa6a5bbd74d22916918a4

The execution service checks the pinned package files before running. The original bundle retains its source metadata, including historical publication and user-testing fields; those fields do not describe this website.

Source & scope

What this run can show

A fresh run reproduces the eight-case HCRM component of the ME-World coverage review using the original evaluation code. It does not reproduce the entire coverage study or establish independent validation or user benefit.

This is a bounded prototype for map values, deletion, and additive counters, not the official Automerge implementation. Text editing, storage, synchronization, and production collaboration are outside these cases.

Inspect the eight original cases

Disjoint concurrent edits survive

Run the original bundle

What this run can show

Privacy & Analytics Consent