Menü
Use when working on the ZettelId internals
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
| name | features-zettel_ids |
| description | Use when working on the ZettelId internals --- the Yin/Yang provider, the triangular-number coordinate mapping, the Index interface and its v0/v1 implementations, raw-binary persistence, calls to `CreateZettelId` / `AddZettelId` / `PeekZettelIds`, ZettelId genesis from `dodder init`, or `ErrZettelIdsExhausted` debugging. |
User-facing behavior (CLI format, exhaustion at the command level) lives in the plugin copy of this skill. This file covers the implementation: data structures, source locations, and Index semantics.
A ZettelId is a struct with left and right string fields, rendered as
left/right. Both parts must be non-empty identifiers. Parsed via the
doddish tokenizer (identifier / identifier).
Source: go/internal/bravo/ids/zettel_id.go --- ZettelId struct,
Set(), String().
Each repo stores two newline-delimited word lists created at dodder init
time:
DirObjectId()/YinDirObjectId()/YangDirObjectId() resolves to layout.MakeDirData("object_ids") within the
.dodder repo data directory (not .madder). The files are copied from
-yin and -yang flag paths during genesis.
Total ID space = len(Yin) * len(Yang) combinations. The provider lists
are immutable after creation.
Source: go/internal/echo/zettel_id_provider/ --- Provider struct
with yin/yang fields. Reverse lookup of a string -> coordinate is a
linear scan.
A triangular-number mapping converts each 2D (Left, Right) pair to a
single 1D integer and back. This allows the index to store a flat set of
integers.
n = Left + Right + 1; ext = Extrema(n); return ext.Left + Leftn = round(sqrt(id * 2)); ext = Extrema(n); Left = id - ext.Left; Right = ext.Right - idLeft = (n-1)*n/2 + 1, Right = n*(n+1)/2Source: go/internal/0/coordinates/kennung.go ---
ZettelIdCoordinate{Left, Right uint32}, Id(), SetInt().
The Index interface (go/internal/foxtrot/zettel_id_index/main.go)
tracks which IDs are available:
type Index interface {
errors.Flusher
CreateZettelId() (*ids.ZettelId, error)
interfaces.ResetableWithError
AddZettelId(ids.Id) error
PeekZettelIds(int) ([]*ids.ZettelId, error)
}
The factory MakeIndex is gated behind a hardcoded if true { v1 } else { v0 } switch --- v1 is the currently active implementation.
Uses collections.Bitset indexed by 1D coordinate integer. Persists via
encoding.BinaryMarshaler on the bitset --- not gob, not the v0 uint32
stream.
Source: go/internal/foxtrot/zettel_id_index/v1/main.go.
Uses map[int]bool where each key is a 1D coordinate integer. Presence in
the map = available.
Reset() --- Rebuilds the full pool: iterates all (l, r) pairs
from 0..len(Yin)-1 x 0..len(Yang)-1, converts each to a 1D
coordinate, inserts into AvailableIds.AddZettelId(id) --- Marks an ID as consumed: reverse-looks up the
left/right strings in the providers to get coordinates, converts to 1D,
deletes from AvailableIds.CreateZettelId() --- Allocates a new ID: picks from the available
pool (random or predictable), deletes it, maps the coordinate through
Yin/Yang providers to produce the string ID.PeekZettelIds(n) --- Preview up to n available IDs without
consuming them.Persistence: raw big-endian uint32 stream at FileCacheObjectId()
(resolves to DirDataIndex("object_id")). Layout: uint32 count,
followed by count × uint32 coordinate keys, via binary.Write.
Earlier revisions used gob encoding; the migration off gob and a runtime
sanity check on count were added to prevent multi-GB allocations when
the on-disk format drifted (see the comment referencing
amarbel-llc/dodder#68 in v0/main.go).
Thread safety: sync.Mutex protects all map operations.
Source: go/internal/foxtrot/zettel_id_index/v0/main.go.
Controlled by configCli.UsePredictableZettelIds():
rand.Intn(len(AvailableIds) - 1), then iterates
to the Nth entry.When the available set is empty, CreateZettelId() returns
ErrZettelIdsExhausted.
Source: go/internal/echo/zettel_id_provider/errors.go.
The Store (go/internal/oscar/store/) holds a zettelIdIndex field.
When a zettel is written, AddZettelId() is called to mark it consumed.
CreateZettelId() is called when minting new zettels
(go/internal/sierra/repo_actions/write_new_zettels.go,
checkin_haustoria.go, op_checkin.go in romeo/local_working_copy/).
The index is flushed as part of Store.Flush().
| Concern | Package |
|---|---|
| ZettelId type & parsing | go/internal/bravo/ids/zettel_id.go |
| Coordinate mapping | go/internal/0/coordinates/kennung.go |
| Yin/Yang provider files | go/internal/echo/zettel_id_provider/ |
| Index interface & factory | go/internal/foxtrot/zettel_id_index/main.go |
| v1 index (bitset, active) | go/internal/foxtrot/zettel_id_index/v1/main.go |
| v0 index (map-based, dormant) | go/internal/foxtrot/zettel_id_index/v0/main.go |
| Genesis (Yin/Yang copy) | go/internal/foxtrot/env_repo/genesis.go |
| Directory layout paths | go/internal/bravo/directory_layout/v3.go --- DirObjectId(), FileCacheObjectId() |
| Exhaustion error | go/internal/echo/zettel_id_provider/errors.go |
| Store integration | go/internal/oscar/store/ |
Guides day-to-day dodder and der operations including creating zettels, querying and filtering objects, checking out working copies, organizing content, syncing with remotes, and managing tags, types, and dormant objects. Activated by requests to create, edit, organize, query, show, checkout, checkin, push, pull, sync, tag, or otherwise operate on dodder repositories.
Use when working with object genres, adding genre filters to queries, parsing sigils like :z :t :e :k :b, or modifying genre-based dispatch. Also applies when encountering Genre byte type, genre bitwise operations, MakeGenre, or optimizedQueries maps.
Introduces dodder, a distributed zettelkasten and content-addressable blob store. Covers core concepts (zettels, object IDs, tags, types, blobs, the store), installation via Nix, a first-steps walkthrough including repository initialization and zettel creation, and the mental model for working with dodder. Activated when a user is new to dodder, asks what dodder is, wants to get started, install, set up, or learn how dodder works.
Use when working on the dodder/madder blob-store internals
Use when creating, querying, or sharing dodder blob stores at the command line
Use when interacting with dodder ZettelIds (two-part IDs like `ceroplastes/midtown`), seeding a new repo via `dodder init -yin` / `-yang`, choosing predictable or random ID allocation, or interpreting "zettel ids exhausted" errors at the command line.