# Planned features

Ideas that are deliberately out of scope for the current version, tracked here
(and/or as GitHub issues) rather than built half-finished.

## Codeforces baseline sync hardening

The baseline data (see [README § Codeforces baseline
data](README.md#codeforces-baseline-data)) is now a real daily background
job — `supabase/functions/sync-cf-baseline`, cron-triggered every few
minutes, sampling 1,500 real users per tier — not a manually-regenerated
static file. A few things are deliberately left for later rather than
over-built now:

- No monitoring/alerting if the cron job silently stops advancing (e.g. CF
  changes their API shape, or the function starts erroring every
  invocation) — you'd only notice by seeing `cf_baseline_data.updated_at`
  go stale. Worth a scheduled check that alerts if it hasn't updated in,
  say, 36 hours.
- Failed per-user fetches within a batch are silently skipped (logged, not
  retried) — if Codeforces has a bad moment during a run, that batch's
  sample is just slightly smaller, not wrong, but there's no backfill.
- The sampling pool is rebuilt from a fresh `user.ratedList` call at every
  phase transition (4 times per cycle) rather than cached once — simple and
  self-correcting, but slightly wasteful.
- This is also the natural place to build a real top-CF-users leaderboard
  someday, since it needs the same kind of scheduled server-side job.

## Other ideas

- ~~Real accounts shipped: Codeforces-handle sign-in (Supabase Auth anonymous
  users + `cf-signin` Edge Function) with server-side verification of the
  compile-error submission, one verified handle per account, server-stored,
  synced across every device you sign into (see README's "Account setup").
  A DB-level uniqueness constraint guarantees the same handle can never be
  verified on two accounts. The `cf-signin` Edge Function re-checks the
  compile-error proof server-side, so a determined client can't skip it.~~
  **Done** — shipped in the current version.
- ~~Make sign-in optional instead of a hard gate. Unauthenticated users should be
  able to freely browse the roadmap topics, compare teams, and view Codeforces 
  stats for any public handle temporarily, keeping only progress and points 
  locked behind account verification.~~
  **Done** — shipped.
- Map raw Codeforces tags (`dp`, `binary search`, `number theory`, etc.) to
  the 103 roadmap topics, so weak-tag call-outs (Codeforces Analysis, the
  Team Analyzer's team-gap list) can link straight to "practice this roadmap
  topic" instead of just naming the raw tag. Checked this while building the
  Team Analyzer — no such mapping exists today, and topic `id`/`name`
  strings only cleanly match real CF tags in a handful of cases (most are
  compound/differently-phrased, e.g. topic "Bitmask DP" vs. tag `bitmasks`),
  so it needs real curation across all 103 topics, not a naive string match.
- Contest-type breakdown (Div1/Div2/Div3/Educational/Global performance
  shown separately) — some people do well in practice but underperform live,
  or vice versa.
- Rating milestone projection — a rough trend-line extrapolation, clearly
  caveated as a rough estimate, not a guarantee.
- PWA/offline support (service worker + manifest) so the app installs and
  works without a network connection at all.
- Per-subject custom pacing weights (e.g. let the user say "I already know
  most of Math, weight it lower") instead of pure topic-count proportionality.
- Configurable point-formula constants (base divisor, focus multiplier,
  minimum) via the profile card instead of fixed constants in code.
- Solve-velocity-aware fallback timeline estimate (use actual recent pace
  instead of the fixed default-topics/week assumption) once checklist changes
  are timestamped.
- Optional CSV export of the solved log for spreadsheet users.