Repository navigation
Conversation
rbuergi
added a commit
that referenced
this pull request
May 5, 2026
…p trigger Tasks #15, #16, plus a related production-code drift fix. 1. CodeNodeType.HandleExecuteScript: workspace.UpdateMeshNode that stamps LastExecutedAt / LastExecutedBy / LastActivityPath onto the Code node was being discarded. Observable.Create only runs on Subscribe — the stamp never landed, so CodeNode_StampsLastExecutionFields timed out waiting for LastActivityPath to match. Added .Subscribe(_ => {}, _ => {}). 2. ScriptExecutionInUserHomeTest.Cancel_Via_RequestedStatus_Patch...: the test's workspace.UpdateMeshNode that flips RequestedStatus = Cancelled on the activity log had the same bug — the patch never hit the activity hub, so the kernel never got the cancel signal. Added .Subscribe(_ => {}, _ => {}). 3. NodeTypeStreamCache.MaybeKickCompile: was flipping CompilationStatus = Pending to trigger compile via InstallCompileWatcher — which was removed in 86b3470. The flip is dead code now; nothing reacts to Pending. Replaced with a CreateReleaseRequest post to the per-NodeType hub, matching the explicit-only compile flow. 4. MeshNodeExtensions.UpdateMeshNode<TContent>: the void overload now forwards to GetMeshNodeStream(path).Update(...) and returns IObservable<MeshNode> — callers must Subscribe (the cold observable's side effect only runs on Subscribe, per Doc/Architecture/AsynchronousCalls.md). No callers in src/. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
rbuergi
added a commit
that referenced
this pull request
May 10, 2026
…, parallelism, backoff)
NuGetAssemblyResolver:
- Evict faulted/cancelled tasks from the per-key cache before
returning. A transient feed failure (network, throttle, cancelled
in-flight resolve) used to poison the cache for the resolver's
lifetime — every subsequent call replayed the same exception.
- Pass CancellationToken.None to the shared core task so a single
caller's cancellation can't take down the resolution for
others; per-caller `ct` projects via `task.WaitAsync(ct)`.
- Switch DependencyBehavior from `Lowest` to `HighestMinor` so
`#r` directives pick up patch-level security fixes via
transitive dependencies without silently jumping major/minor.
- Document that hydrated cache content is trusted to match
(id, version) — flag for future content-hash verification if
cache poisoning becomes a concern.
LinkedInPublisher / XPublisher (LinkedIn already committed in batch A
for the dynamic+PII parts; this commit adds the 401 retry):
- SendWith401RetryAsync: on the FIRST 401 response from a publish,
force-refresh the token (zero ExpiresAt before EnsureFreshAsync)
and retry once. Closes the race where the access token's TTL
expired between EnsureFreshAsync and the actual API call.
PostStatsRefresher:
- Process due-refresh targets via Parallel.ForEachAsync bounded
by SocialOptions.StatsRefreshDegreeOfParallelism (default 8),
so a slow API + large refresh window can't let one tick
overshoot the next interval.
- Per-target failure backoff via a ConcurrentDictionary of
last-failure timestamps — targets that failed within
StatsRefreshFailureBackoff (default 15 min) skip the next tick.
Stops a degraded platform from generating thousands of repeat
warnings every cycle while the underlying issue is fixed.
Success clears the backoff entry.
SocialOptions: add StatsRefreshDegreeOfParallelism (8) and
StatsRefreshFailureBackoff (15 min) knobs.
Addresses PR #95 review items #12, #13, #14, #16, #17, #18.
(#15 XPublisher defensive parse + the LinkedIn dynamic / PII items
were already in commit 478fdaa.)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
rbuergi
added a commit
that referenced
this pull request
Jun 5, 2026
…-free top-level autocomplete + email docs #16 — public.top_level_index MATERIALIZED VIEW over every partition's namespace='' root node (one row per partition), built by public.rebuild_top_level_index() from public.searchable_schemas. Powers top-level autocomplete from one small indexed relation instead of a cross-schema fan-out (the connection-pool storm). NOT a node copy; NOT rebuilt on the query hot path (DDL-per-query deadlocked — removed); re-materialized only at schema-init (PostgreSqlSchemaInitializer) and at deploy (SearchableSchemasUpdater). Test: TopLevelIndexTests (materializes partition roots only, excludes within-partition children, idempotent rebuild). Docs: SendingEmail / InvitationOnlyOnboarding / OnboardingNewEnvironment updated to Email__MailboxAddress (renamed from NoReplyAddress) + bidirectional (Mail.Send outbound, Mail.ReadWrite inbound). Also bundles concurrent in-flight branch work present in the tree: the TypeRegistry / MessageHubConfiguration default $type discriminator change (FullName -> simple Name, to fix cross-hub DeliveryFailure) and ThreadExecution / ThreadSubmission edits. Build: 0 errors, 0 warnings. PG tests (TopLevelIndex + PartitionLifecycle): 6/6 green. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
rbuergi
added a commit
that referenced
this pull request
Jun 5, 2026
…view + unscoped query defer #16 matview: rebuild_top_level_index() now selects exactly the partition ROOT per schema (namespace='' AND id=<schema_name>) so path (a generated column = id when namespace='') is globally unique — the prior (namespace,id)/(path) UNIQUE index collided on non-root top-level nodes repeating ids across partitions. atioz migration completes; matview = one row per real partition. #20 (partial): StorageAdapterMeshQueryProvider defers UNSCOPED queries to the native partitioned provider's SQL fan-out (partitioned PG only, via StorageAdapterQueryProviderOptions). Removes the pedestrian's slow cross-partition ListChildPaths walk for the onboarding middleware's unscoped 'nodeType:User content.email:X' lookup (the 20s timeout). Scoped/satellite queries untouched (pedestrian is the real server for _Access reads) — 452/452 PG tests pass. Absent-partition scoped-walk slowness (_Access/User/_UserActivity) is a separate routing-layer fast-fail, tracked next. RoutingConvergenceTests: PG regression guards that ResolvePath/GetQuery converge to empty (no hang) for absent partitions + non-matches. Docs: CLAUDE.md + Deployment.md rewritten to the AKS image-update deploy (dotnet publish -t:PublishContainer → az aks command invoke set image/rollout); deploy.sh = first-time env setup only; bare aspire deploy = legacy. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
rbuergi
added a commit
that referenced
this pull request
Sep 11, 2026
…in on 10.0.12 MeshWeaver.Plugins run 34594554211 (main), MeshWeaver.FutuRe.Test, pid 51566: SEGV_MAPERR at 0x0, background_sweep()+0xa61 on a dedicated BGC thread, the cursor's MethodTable word reads zero; libcoreclr build-id 79945f51 (stock 10.0.12) read out of the crashed process. Trace log complete and clean (44/44 teardowns clean, alc=1 throughout). The second post-fix sighting; grouping, eliminations and cause/victim updated to sixteen. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
rbuergi
added a commit
that referenced
this pull request
Sep 11, 2026
…ery run's runtime read 719 portal-hosts verdict runs (2026-09-06 → 09-11), runtime read per run from its own log: 10.0.11 8/444 = 1.80 % (CI 0.78–3.52), 10.0.12 2/275 = 0.73 % (CI 0.09–2.60); rate ratio 0.40, exact CI 0.04–2.02, p = 0.20. Elimination by counter-example (#15, #16); reduction not established — ~2,030 runs per arm (~2026-09-24). The /sigsegv skill now says so. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
rbuergi
added a commit
that referenced
this pull request
Sep 11, 2026
… collectible contexts REALLY unloaded Unload() only requests an unload; the runtime frees a collectible context over later collections, on the finalizer thread. MeshTeardownSignal fires when teardown has requested them, so the next test's mesh was built while the previous mesh's NodeType contexts were still being torn down — the state every readable FutuRe crash dump (#12-#16) caught: 3-7 NodeAssemblyLoadContexts with _state=Unloading while the next instance ran. - CollectibleContextUnloads (Mesh.Contract, mesh-scoped): records retired contexts by SIGNAL, never by reference; AllCollected completes when every one has been collected, errors when an unload was abandoned (drain fault / Unload() threw), emits synchronously when nothing is pending. - NodeAssemblyLoadContext.RetireInto: a finalizable sentinel referenced only by the context (Unload() suppresses the context's own finalizer); it can only be finalized once the runtime has destroyed the LoaderAllocator and released the context. - CompilationCacheService retires every context it disposes into the tracker. - Test bases (Monolith, Hub): teardown ends with CollectibleUnloadDrain — drive full collections, then observe AllCollected; a fault fails the class, a retention is reported. - Controls: RetiredContextCollectedSignalTest. - Record: DebuggingNativeCrashes + /sigsegv corrected (alc=1, #15 "live referenced", eliminations 1/2/6). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This was referenced Sep 19, 2026
This was referenced Sep 21, 2026
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.
No description provided.