@work.books/substrate
v0.1.0
Published
Workbook substrate v0 — file-as-database persistence (parser, hydrator, integrity guard)
Readme
@work.books/substrate
Workbook substrate v0 — file-as-database persistence for workbook HTML files.
Goal
A workbook is a single .workbook.html file that is the database. State (yjs CRDT bytes for the spec/composition, optionally SQLite bytes for relational data) lives inline in the file. Edits are persisted by writing the file back to disk through the strongest transport available on the user's browser/OS.
File model (v0)
<!DOCTYPE html>
<html>
<head>
<meta name="workbook-substrate" content="v0">
<script type="application/json" id="wb-meta">
{"workbook_id": "uuid", "schema": 0, "runtime_version": "0.1.0"}
</script>
<!-- Cold image. Compacted periodically. One per data container. -->
<script type="application/octet-stream"
id="wb-snapshot:composition"
data-cid="bafy…"
data-format="yjs">base64…</script>
<script type="application/octet-stream"
id="wb-snapshot:data"
data-cid="bafy…"
data-format="sqlite">base64…</script>
<!-- Hot log. Append-only at the format level; physical writes
depend on the transport. -->
<script type="application/json" id="wb-wal">
[
{"seq": 1, "target": "composition", "parent_cid": "bafy…",
"cid": "bafy…", "payload_b64": "…"},
{"seq": 2, "target": "data", "parent_cid": "bafy…",
"cid": "bafy…", "payload_b64": "…"}
]
</script>
<script type="module" id="workbook-runtime">…</script>
</head>
<body>
<div id="app"></div>
</body>
</html>All persisted bytes live inside <script type="application/octet-stream"> or <script type="application/json"> data blocks — parser-inert, content preserved verbatim, retrievable via document.getElementById(id).textContent or via fetch(self.location).then(r => r.text()) for binary-safe extraction.
Status
In development. See spikes/ for in-progress feasibility studies and bd show core-1ja for the project plan.