Holo-web-conductor support - #47
Open
zippy wants to merge 2 commits into
Open
Conversation
zippy
commented
May 8, 2026
zippy
commented
Contributor
- Adds support for running under holo-web-condcutor.
Adds end-to-end support for running emergence under the Holo Web Conductor browser extension, with a single-command local dev stack that spins up bootstrap + conductor + h2hc-linker + joining service + hc-spin progenitor sharing one network seed. Modelled on ../unyt/deploy/local-dev.sh and ../mewsfeed's NetworkInfo connection indicator pattern. Stack (deploy/local-dev.sh + deploy/start-hwc.sh): - start-hwc.sh wraps `npm run start:hwc[:joining]`, exporting one NETWORK_SEED both as itself (read by local-dev.sh) and as VITE_PROGENITOR_NETWORK_SEED (read by App.svelte). Default `emergence-local-dev`; override with `NETWORK_SEED=foo npm run …`. - local-dev.sh repacks workdir/emergence.happ with $NETWORK_SEED stamped into dna.yaml and stages it to ui/public/emergence.happ — required because @holo-host/web-conductor-client's joinAndInstall doesn't pass dna_modifiers.network_seed through to installApp, so the bundle's manifest seed has to match the conductors'. - kitsune2-bootstrap-srv runs both bootstrap and iroh-relay on the same HTTP listener (`no_relay_server: false`); the relay URL = the bootstrap URL. Parsing the QAD port instead leaves iroh stuck without a current url and silently breaks peer discovery. - Linker runs in open-auth mode (no admin secret) even with the joining service, so a stale HWC install can't get locked out by an empty in-memory allowlist after a linker restart. - hc-spin is launched WITHOUT --network-seed; emergence's model is "provisioned cell = empty template, each network is a clone." If hc-spin baked the seed into the provisioned cell, the UI's auto-join would collide on DNA hash and the conductor would return DuplicateCellId. Yarn/npm surface: npm run start:hwc backend + vite + hc-spin (no joining) npm run start:hwc:joining + joining service (recommended) npm run start:hwc:progenitor spawn hc-spin standalone npm run stop:hwc tear down everything npm run status:hwc show what's running UI changes (ui/src/App.svelte + helpers): - isHwcContext() in utils.ts: true on ?runtime=hwc query param OR window.holochain.isWebConductor (matches mewsfeed's runtime-config pattern as a query param so we don't ship a config file with the .webhapp). - New HWC connect branch uses WebConductorAppClient with roleName: "emergence" (HWC's synthetic appInfo defaults to "default"/<app-id> without it, breaking CloneManagerStore). URL params override env vars: ?joiningServiceUrl, ?linkerUrl, ?happBundlePath. happBundlePath defaults to /emergence.happ. - VITE_PROGENITOR_NETWORK_SEED is the non-Moss stand-in for the Weave tool-installer affordance: when set, the UI in non-Weave/ non-HWC mode (i.e. the hc-spin progenitor window) auto-joins a clone with that seed, skipping NetworkOnboarding entirely. join() is wrapped in DuplicateCellId fallback that re-lists clones and activates the existing one — robust against hot-reload + residual state. amSteward = (env var set), so the Admin pane's initial- config wizard auto-shows. Production builds leave the env var unset, so HWC browser users do NOT become stewards. - CloneManagerStore gains an `isHwc` constructor flag; HWC mode uses the provisioned cell directly (the joining service installs with dna_modifiers.network_seed, no clone needed) — same code path as the existing Weave branch. - HwcConnectionStatus.svelte (adapted from mewsfeed/components/ NetworkInfo.vue) subscribes to the HWC client's connection:change events and renders Extension/HTTP/WS/Auth/Linker/Peers/Reconnect/ Errors live, plus a derived boot-stage label so the loading screen shows exactly which step is stuck. Wired into App.svelte's three waiting states: pre-connect spinner, post-connect sync, and init-error. .npmrc: legacy-peer-deps=true. @holo-host/web-conductor-client@0.1.0 declares peer @holochain/client ^0.18.0 || ^0.19.0 || ^0.20.0 and emergence pins ^0.20.4-rc.0 — strict-mode npm rejects the prerelease as unsatisfied; runtime APIs are compatible. .gitignore: linker-key.ed25519 (h2hc-linker writes its identity keypair to cwd on first start). docs/dev/hwc-manual-test.md walks the flow, why .happ is repacked, why bootstrap == relay, the env-var → seed plumbing, and a troubleshooting section keyed on the specific error messages we hit.
…inker Three small follow-ups to the local HWC dev stack from 932576e. clean subcommand ---------------- `stop` deliberately leaves /tmp/emergence-local-dev intact so you can post-mortem linker.log / conductor.log / progenitor.log after a failed session. Wiping that state is now an explicit action: `npm run clean:hwc` (or `./deploy/local-dev.sh clean`). It refuses to run while any tracked PID is alive — you have to `stop` first — and also removes the staged ui/public/emergence.happ so a subsequent start re-stages a freshly seed-stamped bundle. Terminology + log map --------------------- Reframed the script + manual-test doc around "three k2 nodes" instead of "the linker's conductor". The h2hc-linker binary itself is a k2 node; the `hc sandbox` instance it points at via H2HC_LINKER_CONDUCTOR_URL is a separate k2 node ("background conductor"); and hc-spin runs a third k2 node ("progenitor"). All three peer up via the same bootstrap+relay. Stopped using "linker conductor" to refer to either of the latter two — it conflated roles that are independent at the kitsune2 layer. Doc gains a small table mapping each of the three log files to which k2 node wrote it, plus a one-liner on reading `iroh::magicsock me=/remote=` lines to confirm full-mesh peering. Always rebuild the linker on start ---------------------------------- Replaced the "rebuild only if missing or `--help` fails" gate with an unconditional `cargo build --release` against $H2HC_LINKER_DIR. Cargo's own incremental check is the up-to-date oracle (~0.6s no-op when nothing changed), and the prior gate silently shipped stale binaries when ../h2hc-linker source was edited between runs. The stale-nix-interpreter recovery branch was kept and moved to run *after* a successful build, gated on the binary not responding to `--help`. cargo's stdout is suppressed unless the build fails, in which case a verbose re-run command is printed.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.