• v1.0.0-beta.5 3a4d1af2ea

    Morphit v1.0.0-beta.5
    All checks were successful
    morphit-ci / TypeScript typecheck (sweep all workspaces) (push) Successful in 36s
    morphit-ci / apps/web svelte-check (svelte-kit sync + svelte-aware tsc) (push) Successful in 28s
    morphit-ci / ansible-lint (playbook quality gate) (push) Successful in 10s
    morphit-ci / Smoke suite (run-smokes.sh, triple-pulse) (push) Successful in 14m54s
    morphit-release / Build + publish release tarball (push) Successful in 15m49s
    Stable

    agorise released this 2026-06-04 21:07:26 +00:00 | 4 commits to main since this release

    Signed by agorise
    GPG key ID: 53524E1F1017EB9C

    Two operator-facing improvements on top of beta.4. First, your
    instance is now far more resilient to flaky Blurt RPC nodes — a
    single dead or rate-limited node no longer stalls your sync, and
    you can check your endpoints before you rely on them. Second, you
    can now moderate your own instance: review the abuse signals the
    indexer already collects and hide a troublesome account's listings
    — without touching the chain, anyone's funds, or any other
    instance. This release is recommended for all operators.

    Added

    • Instance-local moderation — review flags and block accounts.
      Morphit's indexer already watches for two abuse patterns: accounts
      that review each other suspiciously often (reciprocity rings) and
      accounts that look like the same person behind several names.
      beta.5 turns those signals into something you can act on. Run
      morphit-ops moderation (or pick Moderation from the
      morphit-ops menu) to review the flagged accounts and block any of
      them right there; you can also block or unblock any account by name
      with morphit-ops block <account> [reason] and morphit-ops unblock <account>.

      A block is instance-local. It hides that account's listings
      everywhere your instance shows them — the public orderbook, the
      per-account view, featured slots, the RSS feeds, and the live
      stream — and nothing more. It does not broadcast anything to
      the chain, does not touch anyone's funds, keys, or identity,
      and has no effect on any other Morphit instance. It is fully
      reversible. A blocked person signed in on your instance sees a
      clear, non-alarming banner explaining that their posts are hidden
      on this instance only and remain visible on every other Morphit
      instance, with a link to reach you — which is the whole point of a
      federation: no single instance can censor anyone across it. See
      OPERATIONS.md §6a and RUN-A-MORPHIT-NODE.md §9.1.2.

    • RPC endpoint health — checked before, and visible during, a run.
      morphit-ops doctor and the setup wizard (morphit-ops init) now
      test each Blurt RPC endpoint you have configured — a real chain
      query, not just a DNS lookup — and tell you in plain English
      whether they are all reachable, some are down, or all are dead,
      before you depend on them. (Pass --no-rpc to doctor to skip
      it.) And /v1/health now reports how many of your RPC endpoints are
      currently healthy, with full per-endpoint detail in the verbose
      view, so you can tell at a glance whether a sync problem is an RPC
      problem.

    • The morphit-ops menu now shows your version and pending flags.
      The menu lists your installed version next to Upgrade (and the
      latest available release, when it can reach the release server),
      and the number of unresolved moderation flags next to
      Moderation, so both are visible the moment you open the menu.

    Fixed

    • A single dead or rate-limited RPC node no longer stalls your
      instance.
      Two related fixes. The indexer now ships with the same
      built-in list of working Blurt RPC nodes the relay already had:
      previously the indexer required you to configure endpoints with no
      fallback, so a node set up with a list that later went dead — while
      the relay quietly ran on its own good defaults — could freeze the
      indexer's sync. Both services now fall back to the same vetted
      four-node list when the setting is absent, and the setup wizard
      writes that same list to both. Separately, an RPC node that is up
      but rate-limiting you (HTTP 429) or briefly erroring (502/503/504)
      is now treated as a reason to rotate to the next node and back off,
      instead of surfacing as a hard failure — so a throttling or flaky
      node is routed around automatically.

    • Quieter, clearer RPC logs. The noisy Didn't failover… lines
      the underlying Blurt library printed on every transport hiccup are
      now suppressed. Your real endpoint health is on /v1/health
      instead.

    Everything from beta.4 still applies

    beta.4 added morphit-ops doctor (a read-only "will my node start?"
    check) and fixed an indexer boot crash that happened when a Matrix
    room was set for operator alerts, plus two settings the setup
    wizard was not writing. See RELEASE-NOTES-v1.0.0-beta.4.md for
    details.

    Upgrading

    • If you installed cleanly from a recent release and your node runs,
      just npx morphit-ops upgrade to pick this up (it carries your
      config and keys forward). After upgrading, it is worth running
      morphit-ops doctor once — it now checks your RPC endpoints too.
    • No configuration change is required on your side. If you had
      manually copied RPC endpoints into MORPHIT_INDEXER_RPC_ENDPOINTS
      as a workaround, you can keep them or remove them — the indexer now
      has a safe built-in default either way.

    Verify the download

    sha256sum -c morphit-v1.0.0-beta.5.tar.gz.sha256
    

    Output must say OK before you extract.

    Status

    Pre-launch beta. Not yet recommended for production traffic. The
    canonical public instance is morphit.io. Community operators
    welcome — start at docs/start-here/.

    Downloads