#146

feat: add support for Nix via a flake

s12d ago

← Back to list

01Issue

BugTriagedSwamp CLI

Assigneesstack72

Relationships

#1021 extension update/install reports new version while on-disk pulled files stay old

Opened by mgreten · 7/7/2026

Summary

After updating the pin for a pulled extension, swamp extension update / swamp extension install report the extension as already current at the NEW version, while the files actually on disk under .swamp/pulled-extensions/… remain the OLD version. The only way I found to force the real re-fetch was to delete the cache directories by hand. Filing as a question in case this is expected behavior with a flag I'm missing.

What I saw

On a second machine (Roccinante) syncing a repo whose upstream_extensions.json had been bumped in git:

# upstream_extensions.json (from git) says the new versions:
#   @swamp/s3-datastore  2026.07.02.1
#   @mgreten/github-pr-feed  2026.07.06.1

$ swamp extension install
  → "All extensions up to date."     # but on-disk manifest still 2026.06.30.2 / 2026.05.20.1

$ swamp extension update @swamp/s3-datastore
  → "@swamp/s3-datastore: already up to date (v2026.07.02.1)"   # claims new version…
$ grep '^version:' .swamp/pulled-extensions/@swamp/s3-datastore/manifest.yaml
  version: 2026.06.30.2                                          # …but disk is still old

So the version the CLI reports and the version on disk disagree, and neither install nor update reconciles them — they short-circuit on the lockfile/tracking state without checking (or refreshing) the actual pulled files.

Workaround that worked

rm -rf .swamp/pulled-extensions .swamp/bundles .swamp/datastore-bundles
swamp extension install
  → Installing "@mgreten/github-pr-feed"@"2026.07.06.1"…   # now really re-fetches
# disk manifest now correctly reads 2026.07.02.1 / 2026.07.06.1

After the nuke, the active datastore bundle also correctly contained the new (shard-first) code, so it wasn't just the manifest.

Questions

Is this expected — i.e. is the lockfile treated as authoritative and the on-disk files assumed to match, so a divergence (e.g. from a partial earlier pull, or a git-pulled upstream_extensions.json that outran the local cache) isn't detected?
Is there a supported --force / --refresh on install/update that re-fetches regardless of tracking state? If so, the docs/--help didn't surface it to me and I reached for rm -rf instead.
If not expected: could install/update verify the on-disk manifest version against the pin and re-pull on mismatch, rather than trusting the tracking state?

Minor either way — the workaround is reliable once you know it — but the "reports vNEW while disk is vOLD" divergence is a confusing failure mode (it looks like the update worked when it silently didn't), so wanted it on the record.

Environment

swamp 20260707.035329.0-sha.84dbdd93 (both machines)
Multi-repo shared MinIO datastore, per-repo namespaces
Extensions: @swamp/s3-datastore, @mgreten/github-pr-feed

Thanks!

02Bog Flow

Triaged

7/7/2026, 9:53:53 PM

Click a lifecycle step above to view its details.

03Sludge Pulse

stack72 assigned stack727/7/2026, 9:49:48 PM

Scorecard: beta/rc-only extensions show "not scored" despite successful scoring (GET /score resolves only latestVersion)

`swamp extension search --channel beta|rc|stable` ignores the filter — all channels + the no-flag default return an identical result set

Beta-channel extensions are never scored — perpetual 'Scoring is taking longer than expected'

feat: allow models to call other models mid-execution via context.runModel()

swamp-club: enable OAuth provider for swamp serve collective-based admission

extension update/install reports new version while on-disk pulled files stay old

command/shell: step with timeout + a surviving child process reports exit -1 despite command exiting 0

swamp serve startup/shutdown is very slow on large repo and lacks diagnostics

Vault push log overstates what left the machine: Pushed N vault file(s) to datastore counts every dirty file, not just vault files

serve-auth: mode: oauth — device grant client + collective-based admission

serve-auth: enable OIDC Provider + Device Authorization plugins in swamp-club

Orphaned data (migrated-away/deleted model) can't be GC'd or deleted

Docs: document swamp data prune (orphaned-data reclamation)

UAT: CLI coverage for swamp data prune (orphaned-data reclamation)

Session credential revocation scans entire credential map on every dispatch completion

Docs: update worker-commands.md for --all flag and JSON output change

datastore migrate-index aborts on hardcoded 30s S3 requestTimeout for large shards

swamp worker list: hide disconnected workers by default, add --all flag

swamp serve should survive bad extension failures

Worker fleets phase 7: how-to guides and explanation docs

swamp serve: catch unhandled promise rejections from extensions

[@adam/cfgmgmt] SSH helper can crash swamp serve via unhandled finally() rejection

Docs: add --server-token flag to worker connect reference

Add --server-token to worker connect for authenticated orchestrators

Docs: document --trusted-hosts flag for swamp serve

Document SWAMP_SERVE_EXTRA_HEADERS for reverse proxy/tunnel setups

Add --trusted-hosts flag to swamp serve for Docker/Kubernetes worker connections

Allow swamp clients to send custom headers when connecting to swamp serve

Docs: update serve-flags.md default queue timeout from 60s to 10m

Worker fleets phase 6: raise default queue timeout to 10 minutes

Worker fleets phase 4b: capacity slots and concurrent dispatch

MethodContext docs/comments mention context.putSecret but runtime exposes vaultService.put

Docs: add --concurrency flag and SLOTS column to worker-commands reference

install.sh still points users to systeminit/swamp

Support user systemd units for 'swamp serve daemon enable'

swamp serve scheduled execution runs steps in an isolated data scope: readResource sees empty state, writeResource never reaches the model-scope chain — stateful scheduled workflows duplicate instead of deduplicating

Support installing swamp via mise GitHub backend

Worker fleets phase 4a: dispatch runner extraction — child process per dispatch

Docs: add swamp worker daemon enable/disable/status to worker-commands reference

Worker fleets phase 3 PR2: swamp worker daemon enable

Docs: add drain, lifecycle policies, and exit-code contract to references

Worker fleets phase 3 PR1: drain state machine, lifecycle policies, signal handling

Leaderboard search excludes ghosts — a visible ghost can't be found by its alias

issue-lifecycle: plan lifecycle entry silently dropped when summary exceeds 1000 chars

Docs: add worker verify, --verify-on-enroll, and env-var config to references

Labs - Filters do not apply to Blocked status and Billing Tier category

Worker fleets phase 2 PR2: env-var config, swamp worker verify, --verify-on-enroll

Docs: add --type flag to issue edit reference and how-to

Docs: update manual with whole-field expression syntax for workflow task inputs/globalArgs

Allow changing an existing issue to security/confidential after submission

Allow dynamic object expressions for workflow task inputs/globalArgs

Docs: update enrollment token references for fleet tokens (--max-enrollments)

Worker fleets phase 2 PR1: fleet tokens (--max-enrollments, bindings, worker naming)

$identify events are never consumed — anonymous web history never binds to the account

Worker fleets phase 1 PR3: boot reconciliation and integration test

Login success still shows login tip

Support confidential visibility for non-security issues

Fine-grained scopes for swamp.club access tokens

Add associative info to user profiles: collective membership + collaborative contributions

Remove comments from /feed and redesign the page

feed_post_score_events queue grows unboundedly (no TTL, no delete-on-delivery)

Make paid subscribers visually stand out (badge, profile popup summary)

Relax the profanity filter

Giphy integration via /giphy command

@-mention autocomplete for users when writing lab ripples

Worker fleets phase 1 PR2: step_queued events, pending-dispatch model, queue introspection

Docs: add swamp worker queue to worker-commands reference

/lab shipped list is sorted by issue number, not ship date

Docs: update manual for ephemeral data and step-output dependency deferral

Docs: update remote execution references for elastic queueing

Support dynamic host discovery from external sources

Worker fleets phase 1: elastic queueing

catalog backfill should support scoped population for targeted lookups

Docs: document global skill sync behavior for custom tools in manual

swamp update syncs global skills to default agent directories instead of configured custom Pi skillsDir

coerceToSuffix does not sanitize path-unsafe characters, causing forEach items and vary dimensions with / to produce invalid data instance names

Badge for activity on other people's issues

swamp serve: add @keychain=<service>/<account> indirection for secret args on macOS

swamp serve: eliminate wrapper scripts via @vault= secret refs and native startup hooks

Automate mechanical frontmatter fixes in the fixing stage