Record GNPL re-scope: narration/projection, core-factor/gql superseded

The migration record and the live command sheet still describe GNPL as a rename
of the "Glyph query language", extracted from two paths. Both are now wrong in
ways an agent could act on.

Measured, not inferred:

* `lithoglyph/gql-dt/` HAS been extracted -- `hyperpolymath/gnpl` `102c79e`
  (PR #4). MOVED.md still filed gnpl under "migration pending".
* `lithoglyph/core-factor/gql/` has NOT been extracted and will not be. It is
  5 Factor files (gql, lexer-tests, seam-tests, storage-backend, benchmarks)
  and `gnpl` contains zero `.factor` files.

That second gap is now a settled decision rather than an omission. GNPL has
been re-scoped as a narration/projection language -- "what account does this
evidence support, told from whose stance, with what warrant" -- built as a
layer on top of GQLdt and lowering to it:

    GNPL --lowers to--> GQLdt --FFI--> Lithoglyph

`core-factor/gql/` was the runtime/dynamic counterpart to GQLdt's compile-time
proofs. A typed narration layer lowering to GQLdt supersedes it, so it stays in
the monorepo as legacy.

Changes:

* `lithoglyph/MOVED.md` -- splits extracted from not-extracted; records that
  the gql-dt copy here is no longer canonical; states the core-factor/gql
  supersession and why.
* `REGISTRY.adoc` -- splits the old combined row into GQLdt (nested in gnpl,
  canonical, builds) and GNPL (narration/projection). Adds a note recording the
  re-scope and its consequences.
* `RESITE-DATABASES-TO-OWN-REPOS.adoc` -- table row split; the gnpl decision
  marked DONE with the re-scope recorded as a follow-on decision; un-nest order
  corrected.
* `RESITE-CODEX-HANDOFF.adoc` -- **the load-bearing fix.** This is the
  copy-paste command sheet. Step 1d would have run `git filter-repo` over two
  paths and pushed to `hyperpolymath/gnpl`, which is already populated. Marked
  DONE / DO NOT RUN, commands commented out, and the step-0 sanity `ls` no
  longer requires paths that are gone or staying. Step 1e no longer drops
  `core-factor/gql/` from the slimmed parent -- it was superseded, not
  extracted, so it belongs there.

Also notes that the gnpl extract was a squash, so granular history for that
path survives only in the `_split_gnpl` branch, which must not be pruned.

Companion PR: hyperpolymath/gnpl#6.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: hyperpolymath

REGISTRY.adoc

@@ -20,7 +20,7 @@ Resite: decisions finalised; execution per
 |===
 | Database | Repository | Status | Query language (repo)
 | VeriSimDB (cross-modal consistency) | hyperpolymath/verisimdb | exists (canonical) | VCL — nested in verisimdb
-| Lithoglyph (narrative-first) | hyperpolymath/lithoglyph | to create | GNPL — hyperpolymath/gnpl
+| Lithoglyph (narrative-first) | hyperpolymath/lithoglyph | to create | GNPL (narration/projection, over GQLdt) — hyperpolymath/gnpl (exists)
 | Glyphbase (Lithoglyph web UI) | hyperpolymath/glyphbase | to create | —
 | QuandleDB (knot-theory) | hyperpolymath/quandledb | exists (canonical) | KRL — hyperpolymath/krl
 |===
@@ -32,11 +32,37 @@ Resite: decisions finalised; execution per
 | Language | Repository | Status | Notes
 | VCL (Verified Consonance Language) | (in hyperpolymath/verisimdb) | exists | Production; stays nested in verisimdb
 | VCL-UT (usage-tracked extension) | hyperpolymath/vcl-ut | exists (canonical) | Maps to `typeql-experimental/`; reconcile + delete copy
-| GNPL / GQL (Glyph query language) | hyperpolymath/gnpl | to create | From `lithoglyph/gql-dt/` + `lithoglyph/core-factor/gql/`
+| GQLdt (dependently-typed Glyph query language) | (in hyperpolymath/gnpl) | exists (canonical) | Extracted from `lithoglyph/gql-dt/`; `lake build` green. Stays nested in `gnpl` as GNPL's lowering target
+| GNPL (narration/projection language) | hyperpolymath/gnpl | exists (canonical) | *Not* a query language — see note below
 | KRL (Knot query language) | hyperpolymath/krl | exists (canonical) | No KRL content in this monorepo
 | NQC (Nested Query Calculus) | hyperpolymath/nqc | to create | Already standalone in `nqc/` (Gleam)
 |===
 
+[NOTE]
+====
+*GNPL was re-scoped, 2026-07-21.* This registry previously recorded it as
+"GNPL / GQL (Glyph query language)" — i.e. a rename of GQL/GQL-dt, sourced from
+`lithoglyph/gql-dt/` *plus* `lithoglyph/core-factor/gql/`. That reading is superseded.
+
+GNPL is a *narration/projection* language: it answers "what account does this evidence
+support, told from whose stance, with what warrant", not "what is true in the store".
+It is being built as a layer *on top of* GQLdt in the `gnpl` repo and lowers to it:
+
+  GNPL --lowers to--> GQLdt --FFI--> Lithoglyph
+
+Consequences for this registry:
+
+* `lithoglyph/gql-dt/` was extracted to `gnpl` and is no longer canonical here.
+* `lithoglyph/core-factor/gql/` (5 Factor files) was *never* migrated and will not be.
+  It was the runtime/dynamic counterpart to GQLdt's compile-time proofs, and a typed
+  narration layer lowering to GQLdt supersedes it. It stays here as legacy.
+* `docs/migration/RESITE-DATABASES-TO-OWN-REPOS.adoc` is amended accordingly; see
+  `lithoglyph/MOVED.md`.
+
+Rationale: https://github.com/hyperpolymath/gnpl/blob/main/docs/THEORY.adoc (the general
+case) and `docs/LITHOGLYPH.adoc` (what it gives Lithoglyph specifically).
+====
+
 == VeriSim satellites
 
 [cols="1,2",options=header]

docs/migration/RESITE-CODEX-HANDOFF.adoc

@@ -36,7 +36,8 @@ If any check fails, STOP and report exactly which access is missing. Do not work
 | `nqc/` | hyperpolymath/nqc | create | extract as-is
 | `verisim-core/` | hyperpolymath/VerisimCore.jl | create | extract (Julia package)
 | `lithoglyph/glyphbase/` | hyperpolymath/glyphbase | create | un-nest
-| `lithoglyph/gql-dt/` + `lithoglyph/core-factor/gql/` | hyperpolymath/gnpl | create | un-nest (2 paths)
+| `lithoglyph/gql-dt/` | hyperpolymath/gnpl | *exists, populated* | *DONE — do not re-run*
+| `lithoglyph/core-factor/gql/` | — | n/a | *superseded — do NOT migrate*
 | `lithoglyph/` (minus the two above) | hyperpolymath/lithoglyph | create | slimmed parent
 | `typeql-experimental/` | hyperpolymath/vcl-ut | exists | reconcile -> delete copy
 | `quandledb/` (stub) | hyperpolymath/quandledb | exists | reconcile -> delete copy
@@ -53,7 +54,8 @@ gh auth status
 git clone https://github.com/hyperpolymath/nextgen-databases /tmp/ngdb
 cd /tmp/ngdb && git rev-parse HEAD          # record baseline SHA in your final report
 # sanity-check the un-nest subpaths exist before extracting:
-ls -d lithoglyph/glyphbase lithoglyph/gql-dt lithoglyph/core-factor/gql
+ls -d lithoglyph/glyphbase          # gql-dt/ is already extracted (step 1d is DONE);
+                                    # core-factor/gql/ is superseded and stays put
 ----
 
 == Step 1 — create + populate NEW repos (no risk to existing)
@@ -93,15 +95,30 @@ extract lithoglyph/glyphbase/ glyphbase
 publish /tmp/glyphbase hyperpolymath/glyphbase "Glyphbase — Lithoglyph web UI / data layer"
 ----
 
-=== 1d. gnpl (un-nest, two source paths)
+=== 1d. gnpl — *DONE, DO NOT RUN*
+
+[WARNING]
+====
+`hyperpolymath/gnpl` *already exists and is populated* — `lithoglyph/gql-dt/` landed as
+squash-import `102c79e` (PR #4). Running the commands below would push over live content.
+They are retained only as a record of what was executed.
+
+Two amendments to what was originally planned:
+
+. Only `gql-dt/` moved. `core-factor/gql/` is *superseded, not migrated* — GNPL is a
+  narration/projection language layered over GQLdt, which supersedes the Factor runtime
+  counterpart. The Factor sources stay in the monorepo as legacy.
+. The extract was a *squash*, so granular history for this path lives only in the
+  `_split_gnpl` branch. Do not prune it.
+====
+
 [source,sh]
 ----
-rm -rf /tmp/gnpl && git clone --no-local /tmp/ngdb /tmp/gnpl
-( cd /tmp/gnpl && git filter-repo \
-    --path lithoglyph/gql-dt/ --path lithoglyph/core-factor/gql/ \
-    --path-rename lithoglyph/gql-dt/:gql-dt/ \
-    --path-rename lithoglyph/core-factor/gql/:core-factor-gql/ )
-publish /tmp/gnpl hyperpolymath/gnpl "GNPL — Glyph query language (dependently-typed)"
+# HISTORICAL — already executed. Do not re-run.
+# rm -rf /tmp/gnpl && git clone --no-local /tmp/ngdb /tmp/gnpl
+# ( cd /tmp/gnpl && git filter-repo \
+#     --path lithoglyph/gql-dt/ --path-rename lithoglyph/gql-dt/:gql-dt/ )
+# publish /tmp/gnpl hyperpolymath/gnpl "GNPL — Lithoglyph's narration/projection language"
 ----
 
 === 1e. lithoglyph (slimmed: drop the un-nested subdirs; two filter-repo passes)
@@ -110,7 +127,9 @@ publish /tmp/gnpl hyperpolymath/gnpl "GNPL — Glyph query language (dependently
 rm -rf /tmp/lithoglyph && git clone --no-local /tmp/ngdb /tmp/lithoglyph
 ( cd /tmp/lithoglyph \
     && git filter-repo --path lithoglyph/ --path-rename lithoglyph/: \
-    && git filter-repo --invert-paths --path glyphbase/ --path gql-dt/ --path core-factor/gql/ )
+    && git filter-repo --invert-paths --path glyphbase/ --path gql-dt/ )
+# NOTE: core-factor/gql/ is deliberately NOT dropped — it was superseded rather than
+# extracted, so it belongs to the slimmed lithoglyph parent as legacy.
 publish /tmp/lithoglyph hyperpolymath/lithoglyph "Lithoglyph — narrative-first, reversible, audit-grade database"
 ----
 

docs/migration/RESITE-DATABASES-TO-OWN-REPOS.adoc

@@ -23,7 +23,8 @@ Authoritative table: `REGISTRY.adoc`. Summary:
 | `nqc/` | hyperpolymath/nqc | create | extract as-is (already standalone)
 | `verisim-core/` | hyperpolymath/VerisimCore.jl | create | extract (Julia package)
 | `lithoglyph/glyphbase/` | hyperpolymath/glyphbase | create | un-nest
-| `lithoglyph/gql-dt/` (+ `core-factor/gql/`) | hyperpolymath/gnpl | create | un-nest (two paths)
+| `lithoglyph/gql-dt/` | hyperpolymath/gnpl | *done* | un-nest — landed as `102c79e` (PR #4)
+| `lithoglyph/core-factor/gql/` | — | n/a | *superseded, stays here as legacy* (see note)
 | `lithoglyph/` (minus the two above) | hyperpolymath/lithoglyph | create | extract slimmed parent
 | `typeql-experimental/` | hyperpolymath/vcl-ut | yes | reconcile -> delete copy
 | `quandledb/` (stub) | hyperpolymath/quandledb | yes | reconcile -> delete copy
@@ -37,7 +38,14 @@ Authoritative table: `REGISTRY.adoc`. Summary:
 . `verisim-core` -> own repo `VerisimCore.jl` (Julia-package pattern).
 . `verisim-modular-experiment` -> folded into `verisimdb` as a standard feature (no longer
   an experiment).
-. `gnpl` -> un-nest GQL/GQL-DT now into its own repo.
+. `gnpl` -> un-nest GQL/GQL-DT now into its own repo. *DONE* (`102c79e`, PR #4).
+. *`gnpl` re-scoped, 2026-07-21.* This plan originally treated GNPL as a rename of the
+  "Glyph query language", sourced from two paths. Superseded: GNPL is a
+  *narration/projection* language ("what account does this evidence support") built as a
+  layer on top of GQLdt, which it lowers to. Consequently `core-factor/gql/` — the
+  runtime/dynamic counterpart to GQLdt's compile-time proofs — is *superseded rather than
+  migrated* and stays here as legacy. Only `gql-dt/` moved. See
+  `lithoglyph/MOVED.md`, `REGISTRY.adoc`, and `docs/THEORY.adoc` in the `gnpl` repo.
 . New repos: public (org convention). Existing repos (`verisimdb`, `quandledb`, `vcl-ut`,
   `krl`) are populated/canonical -> reconcile, never overwrite.
 . Thinned repo references: hybrid -- submodules for `verisimdb` + `lithoglyph`, registry
@@ -67,8 +75,9 @@ Two patterns:
   branch + PR only if there is unique work. *Never* force-push `main`. Delete the copy here
   only after the canonical is confirmed to hold the content.
 
-Un-nest order for Lithoglyph: extract `glyphbase` and `gnpl` (from `gql-dt/` +
-`core-factor/gql/`) first, then the slimmed `lithoglyph/`.
+Un-nest order for Lithoglyph: extract `glyphbase` and `gnpl` (from `gql-dt/` only —
+`core-factor/gql/` is superseded, see Decisions) first, then the slimmed `lithoglyph/`.
+`gnpl` is done; `glyphbase` remains.
 
 == Blockers
 . *GitHub scope*: the executing session must have access to the destination repos

lithoglyph/MOVED.md

@@ -1,16 +1,35 @@
 // SPDX-License-Identifier: CC-BY-SA-4.0
 // SPDX-FileCopyrightText: 2026 Jonathan D.A. Jewell <j.d.a.jewell@open.ac.uk>
-= Lithoglyph — extraction PLANNED (not yet moved)
+= Lithoglyph — extraction PARTIALLY DONE
 
-The Lithoglyph stack still lives *here*, in `nextgen-databases/lithoglyph/`. Treat
-this directory as canonical until migration completes.
+The Lithoglyph database stack still lives *here*, in `nextgen-databases/lithoglyph/`.
+Treat this directory as canonical for everything not listed as extracted below.
 
-Planned standalone repositories (reserved 2026-06-22; migration pending):
+== Extracted
 
-* Database core -> https://github.com/hyperpolymath/lithoglyphdb
-* Query language (GNPL / GQL-dt) -> https://github.com/hyperpolymath/gnpl
-* Airtable-mode delivery -> https://github.com/hyperpolymath/glyphbase
+* Query language: `lithoglyph/gql-dt/` -> https://github.com/hyperpolymath/gnpl
+  (squash-import `102c79e`, PR #4). *The copy here is no longer canonical.*
 
-CAUTION: a previous version of this file claimed the code had moved to a `lith`
-monorepo. That was INCORRECT — no such consolidation was authorised and no
-`lith` repo exists. The move has not happened yet.
+== Not extracted (still canonical here)
+
+* Database core -> https://github.com/hyperpolymath/lithoglyphdb (reserved 2026-06-22;
+  migration pending)
+* Airtable-mode delivery -> https://github.com/hyperpolymath/glyphbase (reserved
+  2026-06-22; migration pending)
+
+== `core-factor/gql/` — superseded, will NOT be migrated
+
+`RESITE-DATABASES-TO-OWN-REPOS.adoc` originally specified `gnpl` as `gql-dt/` *plus*
+`core-factor/gql/` (5 Factor files: `gql.factor`, `lexer-tests.factor`,
+`seam-tests.factor`, `storage-backend.factor`, `benchmarks.factor`). Only the first
+path was ever moved — `gnpl` contains zero `.factor` files.
+
+That is now the settled position, not an omission. `core-factor/gql/` was the
+runtime/dynamic counterpart to GQL-dt's compile-time proofs. GNPL has since been
+re-scoped as a *narration/projection* language that lowers to GQL-dt, which supersedes
+the Factor implementation rather than migrating it. The Factor sources stay here as
+legacy.
+
+CAUTION: an earlier version of this file claimed the code had moved to a `lith`
+monorepo. That was INCORRECT — no such consolidation was authorised and no `lith`
+repo exists.