Skip to content

Edge edge-20261022-7fff9df

Pre-release
Pre-release

Choose a tag to compare

@github-actions github-actions released this 22 Oct 05:28
· 194 commits to main since this release
7fff9df

Edge build of 7fff9df from main, published as edge-20261022-7fff9df. Edge builds are prereleases built once a day from a commit on main whose CI passed; they do not receive the qualification a stable release does. The latest stable release is v0.13.0.

Compatibility

This build v0.13.0
Version 0.14.0-edge+7fff9df 0.13.0
Storage format written 16 14
Storage formats served v14 to v16 v14 to v14
HTTP contract 0.14 not reported

Storage. A branch this build writes to is left at storage format 16, which v0.13.0 cannot open. Test on a copy of the graph, or back it up, before pointing this build at data that v0.13.0 must still read.

HTTP. This build speaks HTTP contract 0.14. v0.13.0 does not report its contract; its release notes name it.

Install this build

curl -fsSL https://raw.githubusercontent.com/ModernRelay/omnigraph/main/scripts/install.sh | VERSION=edge-20261022-7fff9df bash
iwr -UseBasicParsing https://raw.githubusercontent.com/ModernRelay/omnigraph/main/scripts/install.ps1 -OutFile install.ps1
powershell -NoProfile -ExecutionPolicy Bypass -File .\install.ps1 -Version edge-20261022-7fff9df

Every archive carries a build provenance attestation: gh attestation verify <archive> --repo ModernRelay/omnigraph checks that this repository's release workflow built it from 7fff9df.

Unreleased changes

Release notes recorded since the last stable release, as v0.14.0 will publish them.

Upgrade actions

  • Branch names are used exactly as written. A name with leading or trailing
    whitespace or a control character is now refused as a bad request wherever
    a branch is named (HTTP routes, CLI commands and embedded engine calls), as
    branch statements already refused it; such a name used to be trimmed, so
    main named main. Clients that send padded names must send the exact
    name. A policy whose protected_branches lists such a name now fails
    validation. See branch names.
  • CLI exits follow the disposition. For the data commands the CLI classifies (reads, data writes and aliases), a failure whose effects are unknown, or partial because an earlier step published, exits 2 instead of 1; one with no effects still exits 1, a busy refusal 75 and a failed condition 4, and a server refusal stating effects: none with retry now also exits 75. An embedded write's own refusal now carries the engine's disposition, as a server's does, so embedded load conflicts and failed conditions exit as served ones (1 and 4) instead of claiming unknown effects. A command line the argument parser refuses exits 64 instead of 2 (an addressing refusal inside a command still exits 1), and a failure whose diagnostic cannot be delivered keeps its exit. Maintenance commands, cluster, policy, target, credential and managed commands keep their own exits for now. Read commands, alias included, report the same failure envelope as data writes, with disposition (always effects: none) on stdout in JSON mode and on stderr in human mode. Managed commands keep their own exits for now. Update scripts that read exit 2 as a usage error or exit 1 as every failure. See failed commands.
  • Create profiles explicitly with login --create-profile NAME; ordinary login and managed target discovery honor defaults.profile. Standalone targets now use one flat auth provider: remove named auth maps and default_auth, use separate targets for different identities, and reinstall their stored tokens with credentials set --target NAME --stdin. The old --register, --auth, target register, and config rename-auth spellings are removed. Target listing now shows destinations, managed profiles, and the default marker. See operator configuration.
  • The CLI now selects one complete backend with --target NAME|ADDRESS or defaults.target. Legacy routing flags, profiles, folder selection and use are removed; --config supplies source files only, and resolved targets and known write scopes are always printed to stderr. Cloud bindings preserve native authentication while pinning destination, signing and conditional-write settings; unsupported ambient transport overrides refuse. Run omnigraph target import for legacy operator configuration and lifecycle custody, then review the complete target and endpoint-bound credentials before use. See the CLI reference.
  • The CLI no longer reads OMNIGRAPH_BEARER_TOKEN. It used to send that
    token to any remote server a command addressed, including a literal
    --server https://… URL the operator does not control. A token now reaches
    only the server it is keyed to: a request to a URL under a server declared
    in ~/.omnigraph/config.yaml carries that server's token, and any other
    request carries none. When such a request gets HTTP 401, the CLI names the
    server to declare or the omnigraph login command to run; --json output
    carries this as missing_credential beside the server's unchanged error.
    The server's own OMNIGRAPH_SERVER_BEARER_TOKEN* settings are unchanged.

    If you export OMNIGRAPH_BEARER_TOKEN for the CLI, declare the server under
    servers: and bind its token with omnigraph login <name>, or set
    OMNIGRAPH_TOKEN_<NAME> (the name uppercased, - written as _). In CI,
    write a config.yaml with the server's url into $OMNIGRAPH_HOME and
    pass the secret as OMNIGRAPH_TOKEN_<NAME>. Address the server by name
    (--server <name>) or by a URL under its url. Managed credentials
    (omnigraph login --api, OMNIGRAPH_CONTROL_TOKEN) are unchanged. See
    server credentials.

  • Saved query plans now use envelope version 9, which records each correlated block's strategy. Regenerate plans saved with earlier binaries; older binaries reject version 9 plans.

  • Saved query plans now use envelope version 7 to preserve distinct-count semantics. Regenerate plans saved with earlier binaries; older binaries reject version 7 plans.

  • Vectors written by an earlier omnigraph embed embed the templated text
    rather than the source value and match text queries poorly. Regenerate them
    with omnigraph embed --reembed-all and load the result again. An embedding
    spec, including a seed manifest's embeddings entry, must now list exactly
    one property in fields, the source its @embed declares; a spec that
    lists several is refused, and so is a source value that is not a string.
    See the offline file pipeline.

  • Every error code names the statuses it is sent under. A Blob If-Match that fails
    answers 412 precondition_failed instead of conflict, and 408, 413,
    415 and 416 answer request_timeout, payload_too_large,
    unsupported_media_type and range_not_satisfiable instead of
    bad_request, a resource limit's 413 included. A write the server could
    not settle answers 500 outcome_unknown whatever failed underneath, and
    every 503 that asks for a retry, a graph that is not served or closed
    admission, carries Retry-After as a 429 does. ErrorCode in
    openapi.json names each code's statuses and meaning, two for
    branch_start_refusal (409, and 410 for a commit that is no longer
    retained), every protected operation documents the 404 and 405 its
    router answers, and header
    parameters are published as Omnigraph-Http-Api and
    Omnigraph-If-Actor. Update clients that matched those refusals by
    conflict or bad_request, and regenerate clients that key on header
    parameter names. See HTTP errors.
  • Every error says what it did and what to do next. Server error bodies
    carry a disposition object computed by the server for the request it
    answered. effects is none, partial when an earlier step of the same
    operation published and the failing step was refused before it wrote
    anything (a load that forked its missing branch), or unknown. action is
    fix for a request the caller must correct (its credential, input, path or
    client), refresh for a refusal by the state it met, retry, recover or
    reconcile. A write whose completion the server could not settle answers
    code: outcome_unknown instead of internal; a failure the engine proved
    published nothing keeps its code, with effects: none, or partial after an
    earlier step published. Errors that had no code now
    have one: precondition_failed, recovery_required, external_blob_source
    and change_feed_gap, beside their details, and a deployment status read
    that raced an owner change answers service_unavailable. Refusals that
    answered in plain text now answer the same body: an unknown route (404), a
    method the route does not serve (405), and a malformed body, query string
    or path parameter, where a JSON body that failed to deserialize answered
    422 and now answers 400. CLI data-write failures print the same
    disposition field in place of command_outcome, whose execution is gone
    (not_started is effects: none), taking the server's when its request was
    the command's only effectful one. Update clients that read command_outcome,
    or that told those errors apart by a missing code or a 422. In
    openapi.json, error, code and disposition are required on every error
    body, and the optional query parameters of GET /snapshot, GET /commits
    and POST /load/ndjson are declared as their plain type (string, integer
    or LoadMode) instead of a union with null; they stay optional, so
    regenerate clients built from the earlier description. See
    failed data-write commands.
  • Consumers of saved explain documents must accept explain_version 5 and its new pipeline shape: pipelines entries now carry identity, a pipeline id and depends_on naming the build pipelines drained before it starts, with source, operators and sink as objects whose id joins the physical_plan node ids. The push executor builds from the same decomposition the document serializes, so the printed pipelines are the executed ones. The additive logical_plan_initial field holds the logical plan as resolved, before rewrite passes ran, and explain output prints it as the logical_initial tree.

  • A full-text call (search, fuzzy, match_text, bm25) on a String
    property the schema does not index is refused at type checking (T27).
    Such a call used to scan with a bare, case-sensitive tokenizer, so
    search($d.notes, "deep") missed "Deep Learning" while the same call on
    an @index property matched both spellings. Declare @index on the
    property; every write then records the index's analyzer. An enum takes no
    full-text index: compare its values with =.

    Stored queries are recompiled when a server starts, and a refused one
    quarantines its graph. Run omnigraph queries validate and
    omnigraph cluster plan before upgrading, and treat a T27 finding as a
    pre-upgrade fix. See the search guide.

  • The CLI and server now require exactly Omnigraph-Http-Api: 0.14; 0.13,
    missing, repeated and combined values are refused before graph access or body
    decoding, and the CLI refuses a server that answers discovery with 0.13
    before sending. Contract 0.14 adds the Omnigraph-If-Actor request
    header. Upgrade the CLI, server, Studio and HTTP integrations together. See
    the HTTP contract.
  • --as on a served write is an assertion, no longer a refusal. A request carrying
    Omnigraph-If-Actor: <actor> is refused with 412 actor_mismatch before
    graph lookup or effects when its credential resolves to another actor. --as
    on a served write or a served cluster plan or cluster apply now sends it instead of
    being refused, and a mismatch exits 4 with nothing run; a script that relied on
    that refusal now runs the write when the token's actor matches. The header
    arrives with HTTP contract 0.14, so a server that would ignore it refuses
    the request's contract first. See
    Assert the actor.
  • omnigraph load --json prints the body of the HTTP load route,
    POST /graphs/{id}/load/ndjson, on direct, served and managed targets
    alike. The CLI-only uri field is gone; actor_id, the actor the write is
    attributed to (null when none), is new; and base_branch is always
    present, null without --from. A script that read uri from the load
    output should take the graph address from its own --target, which the
    target line on stderr also reports; the human output still names it. See
    machine-readable positions.
  • Managed login and folder context now use version 2 and select an organization
    explicitly or from a bound folder. Separate organizations keep separate device
    sessions; ambiguous selection refuses, running commands keep the same verified
    identity, and logout clears only the selected organization's control credentials.
    Existing managed users must sign in and bind their folder again. Direct access,
    named-server credentials, graph HTTP contracts and storage formats are unchanged.

  • MCP tool errors carry the HTTP error body. A refused tool call's
    structured content is {"status": <HTTP status>, "error": <error body>},
    the same error, code, disposition and details an HTTP route answers,
    also for the refusals MCP makes itself, which answered {code, message}
    before: an unverified credential is unauthorized, a full endpoint
    too_many_requests, and cancelled, deadline_exceeded and
    result_too_large, which only MCP sends, carry no status. A write that may
    have happened answers code: outcome_unknown with effects: unknown, in
    place of the outcome_unknown: true flag, and a resource read's JSON-RPC
    error carries the same body as data. Deployment routes on a server without
    authentication now refuse with a 403 error body instead of a plain-text
    500, and their bodies keep a 413 or 415 status where they answered
    400. Update MCP clients that read code, message or outcome_unknown at
    the top of an error result. See MCP results and errors.

  • A branch's __manifest now keeps a block registry of its released history:
    one history_block row per released run of a __history block and one
    history_merge row per merge commit inside it, so a cold branch merge
    crosses released history without reading its blocks and reads __history
    only for commits no row covers and for final candidates tied on distance
    and generation.
    Rows beyond the newest 64 live in fragments of __manifest that a publish
    reuses; a push of new registry rows can repack smaller cold fragments. This is
    storage format 15: a format 14 graph is served as it is and each branch is
    stamped 15 by its first publish under this binary; binaries that only read
    format 14 then refuse the graph, so upgrade every binary before the graph. No backfill
    exists or is needed: a merge reads blocks as before until they gain rows.
    The offline omnigraph upgrade converts a standalone format 8, 9 or 13
    graph to 15 and its routes are named history-lance-files-v8-to-v15,
    history-lance-files-v9-to-v15 and history-lance-files-v13-to-v15. See
    the upgrade guide.
  • Operator configuration now uses version: 1 connections, resource targets and owner-local authentication providers, with no backward compatibility or import command. Rebuild old configuration and aliases, use credentials set --stdin instead of static-token login, register remote endpoints, and replace CLI cloud-direct access with a supported access path. Credential status is offline and helper-free; access export is pipe-only. Preserve old operator homes and reconcile unresolved operations with compatible tooling before retiring them; see the operator configuration reference.
  • Managed operator configuration now uses profiles, each with one identity and one auth provider object, instead of connections and nested named auth maps. Rebuild managed references and defaults with profile, use --profile and config rename-profile, and remove managed default_auth and --auth; old forms are rejected without compatibility aliases or an importer. Bare login proposes omnigraph at https://platform.omnigraph.dev, managed targets keep their pinned profile, and standalone named auth remains unchanged; see the operator upgrade guide.
  • New schema declarations interpret @range(p, ..max) as an inclusive
    maximum and reject integer bounds outside I64 instead of rounding them
    to F64. Existing accepted graph schemas retain their stored bounds;
    before reapplying or importing their source, write an old upper-only bound
    as min.. and an oversized integer bound as a decimal with its recorded,
    possibly rounded value. Complete outstanding deployments containing
    affected source before upgrading the binary. See range constraints.
  • A graph whose schema declares @range on a list property now refuses a
    load, mutation or merge that writes a list element outside the range,
    with @range violation; the declaration was not enforced before. Rows
    already stored are left as they are. Writers that relied on the unchecked
    declaration must keep elements inside the range: schema migration refuses
    removing a constraint from a property that stays.

The remaining notes do not fit in a release description; read them in changelog.d at 7fff9df.