Legacy Port

Design: what is left in the old repo

Status: backlog (2026-09-24). An audit of the old enumeratio/enumeratio repo — the Postgres-era predecessor, since deleted; this repo now holds the name — for features that have not come across. The Postgres/pglite plumbing (query engine, service worker, providers, pgdata profiles) is out of scope; what is listed here is mathematics, presentation and UX worth rebuilding in this architecture.

Where the old repo lives

Everything is in ~/Playground/@enumeratio/archive/: enumeratio.git is a bare mirror holding every branch (including the ones that only ever lived in the local checkout), enumeratio.wiki the wiki, and issues.json / pulls.json / project.json the GitHub metadata. Read a file with git -C archive/enumeratio.git show <branch>:<path>. Paths below are on main unless a branch is named.

The catalogue itself came across whole: packages/catalog carries every old collection, statistic, map and crosswalk by name. The gap is behaviour, not names.

1. Representations and table control

The old explorer let one element be seen through several presentations at once, with the printer chosen per column and per scope. None of that layer exists here yet; the closest relative is reduce(expr, env) (rendering-environments.md), which works per cell, not per column or per policy.

• Several glyph kinds per carrier — packages/data/sqlsrc/glyph_kinds.sql: glyph_svg(carrier, kind) with an is_default registry. Permutations drew as matrix, arc diagram, cycle diagram or Rothe diagram; partitions as Ferrers, Young (English/French) or abacus. frontend/src/glyphs.ts has one glyph per carrier and no kind parameter.
• Glyph families not yet drawn — from docs/develop/packages/components/kitchen-sink.md: functional graph (endofunctions), bipartite diagram (surjections), permutation arc diagram (decorated/affine permutations, arrangements), signed bar row, bars with gaps (weak compositions), dot columns (multisets), skew Young diagram (skew/core partitions), polygon dissection, plane-partition grid, sequence bars (ascent sequences, RGS, Gray codes).
• Alternate text notations that round-trip — packages/data/sqlsrc/representations.sql: each named notation (cycle, one-line, blocks, parts, dots, bars, exponential …) is a (render, parse) pair, so any notation is also an input syntax.
• Printer policy — packages/data/sqlsrc/policies.sql: rows scoped to a collection, carrier, category, tag or everything, each with an environment, archetype, clause, mode (permissive, restrictive or override) and priority, resolved by one general→specific fold, GRANT/REVOKE-style. Decides default columns, which printers a column kind allows, and how an environment degrades (print revokes svg and links). Natural fit for the environment capability record.
• Typed columns — packages/client/src/select.ts: a column is one of ~20 kinds (repr, map, stat, dist, pivot, agg, over, glyph, data, title …), each with its own printer set (plain | grouped | katex | link | svg | bars). <notatio-collection-table> columns are untyped expression strings.
• Column controls — packages/explorer/src/ColumnConfigPopover.vue (format, header, min-width, link), PropertiesPane.vue + propRows.ts (reorderable column list, add menu, hide/remove). Same popover reached from the header and from the list.
• Statement bar — StatementBar.vue, PredChips.vue: WHERE / GROUP BY / HAVING / ORDER BY as separately editable, URL-persisted segments, predicates as chips. Group-by switches RowTable.vue into rollup/rowgroup mode; distribution.ts puts a histogram of a statistic over a fibre in a cell.
• Element and identity panes — ElementPane.vue (stepper, stats/maps grid, glyph hero, FindStat-style "which statistic produces these values" solver), IdentityPane.vue (generating function and crosswalk strip), NameThisPanel.vue (match a query against the registry). The crosswalk data is already in packages/reference.
• Unmerged work — feat/composite-printers, auto/explorer-query-model, claude/query-view-policies-spike-6a314f (element relations wired into the query view).
• Never built in either repo — the wiki's Rendering-Printer-Config.md and Render-Assets.md: per-environment printer capability tables, a structured render tree instead of SVG strings (issue #83), link target/history as printer config.

2. Polytopes

packages/symbols/combinatorics/polytope has the permutahedron, associahedron, cross-polytope, simplex and hypercube, and <notatio-polytope> covers what the old WebGL figure did (face selection, face labels from a representation). Left:

• Cyclohedron W_n — packages/data/packs/polytopes/polytope-collections.sql, the largest block. Faces are centrally symmetric dissections of a (2n+2)-gon, the fixed points of the associahedron's half-turn; point and containment reuse the associahedron's Loday point and refinement order; dim = (m/2 − 1) − (orbit count). Faces with k orbits number C(n,k)·C(n+k,k), summing to the central Delannoy numbers (A001850); enumerate by orbits, not by filtering. Issue #330 (open): an explicit Delannoy-path bijection for unrank — try tubings of the cycle graph.
• Vertex-only families — hypersimplex Δ(k,n) on k-subsets (Johnson-graph adjacency), Birkhoff B_n on permutations (permutation matrices, transposition adjacency), type-B permutahedron on signed permutations.
• Overlay — packages/components/src/figures/polytope-overlay.ts: several polytopes in one shared projection (Gram–Schmidt over the union of vertices) to show containment, duality and tiling; demo page docs/develop/playground/helmert-projection.md.
• Guide content — docs/learn/explorations/polytopes.md: duality pairs, majorization (Rado) containment, omnitruncation, tiling by the affine symmetric group, the Tonks projection / sylvester map onto the associahedron, Loday coordinates, the Biane bijection.
• Open question — wiki Element-Relations.md: should a poset's Hasse cover graph and a polytope's 1-skeleton be the same edge data?

3. Combinatorial species

Nothing has come across. In the old repo species were an annotation over hand-written enumerators, not the definition: base_collection_species(collection, species, reading) attached an expression, and base_species_check compared its series against the collection's own cardinality (packages/data/sqlsrc/base_species.sql). The wiki's Species-Data-Model.md proposed inverting that — a collection as a reading of a species, with constructions, generating functions and sequence transforms on one expression tree — but it was never built. Whether to make species the backbone here is an open decision.

• Readings on main — about thirty: permutations E∘C, set partitions E∘E₊, set compositions and surjections L∘E₊, subsets E·E, signed subsets E³, k-subsets E_k·E, words E^k, arrangements E·L, signed permutations E∘C∘(X+X), integer partitions and compositions as isotype readings, Dyck paths / binary / ordered trees 1+X·Y², parking functions, endofunctions, labelled forests and trees as fixpoints (trees via the dissymmetry theorem), Catalan / Motzkin / Schröder as OGF fixpoints.
• Readings on auto/species-info-collections-522f78 (unmerged) — about thirty more, including involutions, derangements, k-coloured and decorated permutations, ballot sequences, Łukasiewicz paths, rook placements, necklaces, Gray codes, the polytope carriers, and the Bell / Fubini / factorial sequences. Also the species_info layer, a Species section on collection reference pages, atoms in the identity pane, and a Species(C) head. See also auto/species-data-orchestration-31d950.
• Machinery — species_kernel.sql (cycle index Z, plethysm, composition), labelled EGF evaluation, the OGF Picard fixpoint solver, species_registry.sql (atoms and operators), species_structures.sql + selfcert-species.mts (a slow definitional enumerator used as an oracle), and element notations like C[{1,2}]C[{3,4}] and E[{E₊[…]}] (species.sql).
• Guide content — docs/learn/explorations/species.md, generic-species.md.

4. Collections without kernels

184 of 282 catalogued collections have no declared head (!ce.lookupDefinition(name) over census's fullEngine()). Of those, 89 are numeric sets (primes, perfect/abundant, figurate families…) — none declared yet, and the first sets the pattern for unbounded collections. The rest, by carrier: permutation classes (Baxter, separable, simple, vexillary, Grassmannian…), binary-word families (bracelets, Stern–Brocot/Calkin–Wilf paths), plane partitions, tableaux (semistandard, skew, shifted, Gelfand–Tsetlin, ASMs), unlabelled/phylogenetic trees, and a long tail of one-offs.

Branches with content not yet here: claude/234-gf (bivariate and recurrence generating- function programs), auto/symmetric-group-slice2-401 (cycle notation, Coxeter-length order), auto/notebook-exact-irrationals (exact algebraic irrationals in output).

5. Docs and design

• Counting guides — docs/learn/guides/ (permutations, stars and bars, subsets and partitions, words and compositions, lattice paths and trees) and docs/learn/explorations/ (bijections, set partitions, subset sum and q-binomials, tableaux, computer science). the guides have no introductory combinatorics.
• Wiki design pages with no counterpart in design/ — Grading, Triangular-Grading, Number-Gradings, Sequence-Transforms, Maps-and-Bijections, Function-Properties, Family-Parameters-and-Numerals (and branch design/family-parameter-tier), Dimensions-and-Measures, Adaptive-Enumeration-Windows, Self-Certification, Element-Relations, Parameterized-Collections, and the surveys (Sage categories, Haskell, Lean mathlib4, Rust traits, browser math libraries).
• Open issues — issues.json has about a hundred open, roughly half Postgres infrastructure. The rest track the design pages above plus sampled non-enumerable collections (real_numbers, complex_numbers). The multicomplex tower issue is done (packages/symbols/algebras/hypercomplex).