Evidence-based investigation across every leak vector (timers, listeners, object
URLs, observers, websockets, static server collections, IMemoryCache, Python
caches). Verdict: no confirmed memory leak — the codebase has disciplined
cleanup. One resource-release correctness bug (over-eager blob-URL revocation in
the CV carousel) was found and fixed (eed9b1f).
Adds docs/performance/: MEMORY_LEAK_REPORT.md, ROOT_CAUSE_ANALYSIS.md,
PERFORMANCE_IMPROVEMENTS.md.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The PDF-carousel cleanup effect had `[pdfCarousel]` deps, so its cleanup ran on
every carousel change and revoked the *previous* array's object URLs — which are
still referenced by unchanged items in the new array. Building a multi-template
deck therefore left every preview except the last with a revoked (broken) blob
URL. Drop paths are already handled explicitly in savePdfToCarousel (replace)
and resetPdfCarousel (clear), so blanket per-change revocation was both harmful
and redundant.
Fix: track the latest carousel in a ref and revoke outstanding URLs only on
unmount (empty-deps effect). No leak either way — unmount still frees them.
Found during the memory-leak/resource audit; this is a resource-release
correctness bug (over-eager revocation), not a leak. profile-page.test: 5/5
green (with an adequate timeout; the suite's 5s-timeout flakiness is pre-existing
and unrelated).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First UI slice toward the product mockups. Shifts the shared MUI theme to match
the mockup identity without restructuring components:
- Default accent -> indigo #6366f1 (drives primary buttons, active nav, and the
Applied/Interview status colours). Users with a custom accent keep theirs.
- Light app background -> #F4F6FB with white paper, giving the layered dashboard
look (cards/inputs sit above a soft grey canvas).
- Global corner radius 8 -> 10 to match the mockups' rounder cards.
- Accent picker: offer indigo/cyan/violet presets first (kept #15803d).
Behaviour-only theme change. Frontend suite: 38/39 pass; the 1 failure
(end-to-end-trust-loop tailored-CV) is pre-existing on this branch (verified by
running it on the clean tree) and unrelated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First Wave 2 (safe refactor) slice. Move the read-only stats/overview aggregation
out of the 3.3k-line JobApplicationsController into a dedicated, injectable
AnalyticsService, and lift its response DTOs (JobStats, FunnelStagePoint,
ResponseRatePoint, CompanyActivityPoint, AnalyticsOverviewDto) into
Models/AnalyticsDtos.cs.
- GetStats: ~44 lines -> 3 (delegates to AnalyticsService.GetStatsAsync).
- GetAnalyticsOverview: ~82 lines -> 3 (delegates to GetAnalyticsOverviewAsync).
- Registered AddScoped<AnalyticsService>(); controller keeps an optional ctor
param with a `?? new AnalyticsService(db)` fallback so the 6 test sites that
construct the controller directly keep compiling.
- Logic is byte-identical (same tenant-scoped context, same projections) so
behaviour is preserved.
Backend suite: 92/92 green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The structured /cv/* calls funnel through a provider router so production can
offload a weak local GPU (GTX 1060) to a cloud provider without any .NET change.
Default stays "ollama" (keyless/local) and /summarize remains local distilbart.
- AI_PROVIDER=ollama|gemini|groq dispatch inside _ollama_generate_json/_text
(entry-point names kept, so no call sites change; Ollama path is byte-identical).
- Gemini (x-goog-api-key header, not URL query) and Groq (OpenAI-compatible
chat/completions) added via stdlib urllib — zero new dependencies.
- /health reports ai_provider + ai_provider_configured.
- Keys read from env only; never logged/committed.
- Compose + .env.example pass AI_PROVIDER/GEMINI_*/GROQ_* through.
Tests: 11 passed (default Ollama unchanged, Gemini/Groq dispatch, missing-key 503,
health reports provider).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add composite indexes (OwnerUserId, IsDeleted) — for the tenant-scoped
list/board/stats/analytics queries that all filter !IsDeleted — and
(OwnerUserId, FollowUpAt) for the reminders surface. Every JobApplication
query is scoped by the OwnerUserId global filter first, so owner-prefixed
composites are the useful shape; the pre-existing single OwnerUserId index
is now a redundant prefix but kept to avoid churn.
Status is intentionally excluded: Pomelo maps the unbounded string column to
MariaDB longtext, which cannot be indexed without a prefix length.
Applied via the startup schema reconciler (StartupInitializationExtensions),
which is how this repo actually provisions schema/indexes on both providers
(SQLite: CREATE INDEX IF NOT EXISTS; MariaDB: MySqlIndexExists-guarded CREATE
INDEX) — NOT via EF migrations, whose committed ModelSnapshot is stale.
OnModelCreating also declares the indexes for model consistency.
Backend suite: 92/92 green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Both endpoints materialised full JobApplication rows (GetAnalyticsOverview also
Include-d full Company) purely to aggregate a few fields, dragging the large
Description/TranslatedDescription/TailoredCvText/Notes/CoverLetter blobs over
the wire on every dashboard load. Project to only the columns each aggregation
needs (mirrors the existing GetTagTrends pattern). Behaviour is identical;
aggregation stays in memory over a small per-tenant set.
Backend suite: 92/92 green.
Note: the planned hot-path *index* migration is deferred — the committed EF
ModelSnapshot is stale (21 lines, no entities), so `migrations add` cannot
produce a clean incremental diff. Resyncing the snapshot is a prerequisite and
is tracked as its own task.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cesnimda
cesnimda