Motifs describe the kind of problem, independent of the material family. One
local tablebase probe returns the root WDL, the DTZ and every legal move with
its own WDL, so each motif is proved without a search or a remote lookup.
- only move: more than one legal move and exactly one preserves the
50-move-aware result, for wins and draws.
- underpromotion: a preserving underpromotion exists while no preserving queen
promotion does.
- 50-move draw: cursed/blessed WDL, or a win whose DTZ exceeds the remaining
halfmove budget.
Motif links are recorded as derived and refreshed as a whole, scoped to the
motif themes, so material, curated and community links survive. Loss endgames
are skipped because only-move is undefined there.
Stalemate is deferred: a stored endgame cannot itself be terminal, so it can
only describe a line and needs a definition before it becomes a predicate.
Release helpers run the derivation without starting the web application, so
`bin/chesstrainer eval` works on the server.
Closes#188
An admin picks a technique theme and sees candidate endgames grouped by Key
with counts. Expanding a family previews its boards without starting a puzzle,
and selecting positions applies the theme in bulk as curated. Large families are
paginated so a 600-endgame group stays usable, and curated links can be removed
from the same surface.
Removal is scoped to curated links, so derived and community links are never
deleted by hand. Material themes are refused because derivation owns them.
Closes#187
Every endgame now belongs to exactly one material family derived from its Key:
pawn, rook, rook and pawn, queen, queen and pawn, minor piece, minor piece and
pawn, basic mate, king and pawn, or bare king other. The bare-king families are
split out because those positions teach a mating or promotion technique rather
than a general material class.
Family membership is objective, so a derivation pass owns links to its themes:
refreshing rewrites them as derived and removes a curated link that contradicts
the family, while technique themes are untouched. Re-running is idempotent.
A mix task applies the mapping to the whole library. Seeds add the king
technique themes for later human tagging.
Closes#186
* add function for upgrading a user's plan in interactive Elixir
* Add theme slugs, kinds and link provenance
Themes gain a stable slug, an optional description and a kind so the app can
tell material, motif and technique labels apart. Slugs are generated from the
name on first save and keep their value when the theme is renamed, so links to
a theme stay valid.
Endgame-to-theme links now record their source (derived, curated or community)
and creation time. Existing links are backfilled as curated. A derived pass can
therefore replace only its own links, leaving human work untouched.
The admin index shows each theme's kind and its coverage across the library.
Creating a theme name requires the manage_themes capability, and members get an
unauthorized error instead, ahead of the suggestion flow.
Closes#185
Selection took a hard rating band (medium = user rating ± 100), removed
everything attempted in the last 30 days, and drew uniformly from what was
left. Because the bulk importer gives every endgame in a batch the same
rating, the band held either the whole library or — once a user's rating
drifted past it — only the endgames that user had already played. With
`fresh` empty, the fallback re-drew from the same band ignoring recency, so
recently played puzzles came straight back. An empty band used a
deterministic nearest-match, alternating between the same two puzzles.
Difficulty is now a bias rather than a filter: candidates are weighted by
`exp(-|rating - target| / spread)` and drawn with an exponential race, so
the whole library stays reachable and a thin region around the user's rating
can never empty the pool. `easy` and `hard` keep one-sided guards so the
labels stay honest. When every candidate is excluded, selection serves the
least recently attempted endgame instead of a random recent one.
Sessions now carry the ids they have served, which fixes the observed
in-session repeat (only the current puzzle used to be excluded), and guest
training gets the same treatment. The 30-day window is configurable as the
repeat window and is now named in GLOSSARY.md, distinct from the rating
window and from the tablebase cooldown.
Difficulty copy is qualitative, since the band numbers no longer describe
how selection works.
* lesson library shown as com,ing soon. re-enable learngin strip
* replace learning strip text with a lasker quote
* remove admin login endpoint
* fix crawler directives and prelaunch login
- noindex the authenticated scope and the 404 catch-all, and cover the
mailing-list confirmation redirect
- emit noindex, nofollow from the robots meta and /health, and assert
public pages stay indexable and followable
- stop inventing lastmod values in the sitemap
- log in through /users/log-in during prelaunch now that the admin login
endpoint is gone; drop create_admin/2 and its orphaned tests
* update caching length to 1 year - name change will invalidate cache
* remove logger
* fix tablebase path breaking tests
* test: check cache headers without requiring built assets
CI's test job does not run mix assets.build, so /assets/css/app.css is a
404 there. Committed piece art is already content-hashed and takes the same
Plug.Static cache option, so the test no longer needs a build step and
passes on a fresh checkout.
* Send transactional email from no-reply@mail.innersquare.app
Centralise the sender identity, reply-to address and SES configuration set in Chesstrainer.Mail so no notifier can drift. Add MAIL_REDIRECT_TO for smoke tests and a mix mail.check task that exercises the SES mailbox simulators, and document the setup in the deploy guide and .env.example.
* Tighten crawler metadata and keep play pages out of search
Canonical URLs no longer carry query strings, public pages get an OG image, WebSite JSON-LD and a per-page language, and the sitemap dates the landing page only. Play and guest-training routes are marked noindex so tens of thousands of near-identical pages stay out of the index.
* Measure built asset sizes with mix assets.sizes
Report raw, gzip and brotli sizes for built JS and CSS, and record the current numbers plus the board-code split recommendation in docs/assets.md.
Tablebase failures now render one message per class, with a Retry action
when fetching again can help. The generic class no longer interpolates
inspect(reason) into the page; the reason goes to the log instead.
Positions outside the tablebase offer the next puzzle (in a session) or the
existing back link, rather than a futile retry.
Train's no-puzzles and exhausted-difficulty flashes become in-page states
with Retry and Change difficulty. The guest pool gets the same treatment,
and GuestLimits.remaining/1 lets a guest see the wall, or the empty pool,
without spending a puzzle.
* change position of theme picker
* Add admin bulk FEN importer
Paste FENs, resolve results from the local Syzygy tablebase only, preview each line's status, and bulk insert with themes and a batch rating.
* Document the remote shell for production user management
* add sitemap to footer links
* remove border under theme selection
* Add admin theme and user management (#144)
Add /admin/themes CRUD and /admin/users list/detail with role and plan editing, guarded by a shared Policy module. Include last-super-admin and self-role protections plus structured id-only admin logs.
Also add admin invitations, confirmation resends and password-reset emails, which work in development through the Swoosh mailbox before SES is configured.
* Rate-limit password logins and emit auth events (#142)
Add a two-bucket login throttle (5 failures per login, 10 per IP in a 15-minute window) and a shared client-address helper for controllers and LiveViews. Blocked attempts skip the database and bcrypt.
Log auth.login.failed and auth.login.throttled warnings with boot-scoped HMAC addresses only, and stop repeated confirmation emails on unconfirmed login attempts.
* Harden the production login edge for Cloudflare
Document the layered login defence: the app throttle is primary, Caddy must pass Cloudflare's CF-Connecting-IP to the app, and Cloudflare edge rate limiting covers request floods.
Update the reference Caddyfile to accept only Cloudflare ranges and reject direct-to-origin traffic, and document the exact Cloudflare dashboard rules.
* Document 10-second Cloudflare rate-limit window
Cloudflare may only offer a 10-second period. Keep the edge rule at 10 login POSTs per 10 seconds per IP and rely on the app's 15-minute login/address throttle for credential guessing.
* Document Cloudflare limits for LiveView registration
Registration submits over the shared LiveView socket, so a Cloudflare rule on POST /users/register will not count account creations. Note the app-level throttle/Turnstile options instead.
* Document current Cloudflare free-plan edge controls
Record the production setup: Cloudflare-only Caddy origins, Bot Fight Mode, and the 10-second login flood rule (10 POSTs per 10 seconds, 10-second block). Clarify that the application's 15-minute throttle remains the real credential lockout.
* Auto-refresh Cloudflare CIDRs for Caddy
Move the origin allowlist into a generated cloudflare-ranges.caddy imported by Caddy. Add an hourly systemd timer and script that fetch Cloudflare's v4/v6 lists, reject malformed responses, validate the candidate Caddyfile, and reload Caddy only when the ranges change. Restore the previous file on validation or reload failure.
* Document Cloudflare CIDR refresher setup and operations
Record the installed paths, permissions, timer checks, manual run, update procedure, and the direct-on-VPS setup fallback when the deploy directory is absent.
* Update README.md
* Add structured JSON production logging (#105)
* Log requests and background failures (#107)
* Document production logging and the VPS runbook
Count distinct users with a session token valid within the 14-day window and use it for the hero players-training stat. The learning strip keeps the total player count.
Also load the split seed modules in seeds_test and drop an invalid 8-piece seed FEN so precommit passes.
* add anonymous guest training with a daily puzzle limit
- /guest-train serves random endgames with no difficulty picker; signed-in
visitors are redirected to /train
- guests get 5 puzzles per UTC day, keyed by an in-memory HMAC of the client
address; the raw address is never stored or logged
- a sign-up wall replaces the next puzzle once the budget is spent
- move the endgame library behind auth so guests cannot browse it
- privacy disclosure tracked in #87
* remove chess-test endpoint and liveview
- /train starts an ephemeral session by choosing easy/medium/hard, banded
relative to the user's current rating (band width configurable)
- session timer, average per-puzzle time, net rating gain and per-puzzle delta
- next puzzle, plus review and retry reused from the play surface
- explicit stop/start: stopping returns to the difficulty picker
- puzzle rating shown once a puzzle completes, and a side-to-move indicator
- extract the play logic/markup into shared PlayCore and PlayComponents so the
standalone play view and sessions share one implementation
- sticky footer; glossary entries for Difficulty and Session; ADR 0009
* remove daisyui and rebuild styles
* very basic setup of new design - missing components etc
* basic navbar with reusable landing components
* basic puzzle of the day
* update todos
* update todos
* add curated landing puzzle cards with selectable board (#55)
* update with footer
* add button to solve today's puzzle
* Make landing page a dead view
Render the home page from PageController with a static board preview, link puzzle cards and CTAs to public-code play routes, count puzzles from the DB, and add a global footer.
* Serve a marketing-only JS bundle on the landing page
Build a small marketing entry point with phoenix_html and the plain-DOM navbar, select it on controller-rendered pages, and keep the full LiveView bundle for LiveView pages.
* Add landing page SEO metadata
Replace the Phoenix default title with a descriptive Lucid Chess title, and add description, canonical, Open Graph, and Twitter card tags to the landing page.
* Complete the hero copy block
Add the spark icon primitive and the Endgames, made clear eyebrow. The primary CTA keeps linking to the active puzzle rather than scrolling to the board.
* Add a persisted global daily puzzle
Assign one endgame per UTC day in a daily_puzzles table with a unique date, preferring endgames unused in the last 30 days and falling back to any. Concurrent first requests insert with on-conflict-ignore and read the winning row. The landing board now shows the daily puzzle.
* remove unused CSS, nav links, time pill
* allow devices to connect to dev server by changing IP address to 0.0.0.0
* Restore the colour-mode switch and add a dark palette
* Use brand colour tokens on the play page
* self host fonts - closes issue #62
* begin splitting CSS into modules
* remove hover on hero chessboard, split more css modules
* remove interactivity from hero chessboard
* update todo
Player-facing endgame links now use a 10-character Crockford base32 public code instead of the database UUID; admin routes keep the UUID. Existing rows are backfilled by migration, and the admin Show page displays and copies the code.
The admin endgame index gains its own search box: exact database id, exact public code, or a FEN / FEN fragment. A single match opens the endgame; multiple matches list below; nothing found says so.
Closes#44Closes#45
- Scaffold password auth (email or username) with phx.gen.auth; no magic links
- Add role (member/moderator/admin/super_admin) and plan (free/premium) enums
- Gate /admin behind admin-or-above via plug + on_mount; hide the nav link
- Add mix accounts.promote for CLI role changes
- Disable email confirmation by confirming accounts at registration
Closes#3
Read the local Syzygy directory from TABLEBASE_PATH so a dev machine and a VPS can point at their own copies. Add mix tablebase.fetch (downloads the 3/4/5 WDL+DTZ set, resumable) and mix tablebase.check (verifies it loads), plus a boot log reporting whether local tablebases are ready and the largest piece count.
Covers the codeable part of #31; installing on the VPS and setting TABLEBASE_PATH there remains a manual step.
Add Backend.Fathom: probes positions with five or fewer pieces via the Fathom NIF, maps WDL/DTZ to the trainer structs (flipping per-move perspective), generates SAN, sorts moves best-first, and falls back to Lichess on any failure. The local directory is configured with TABLEBASE_PATH; local answers are no longer cached.
Add Tablebase.normalize_wdl/1 and flip/1, plus tests for the WDL mapping, SAN, checkmate, fallback, and a cache-free end-to-end local probe.
Vendor the MIT-licensed Fathom C source and compile it into a Rustler NIF (native/fathom_nif). The NIF exposes init/1 (point Fathom at a tablebase directory) and probe/1 (raw WDL/DTZ/move results for a FEN); probes are serialised because Fathom's root probe is not thread safe, and run as dirty NIFs.
Add KRvK/KQvK 3-piece WDL+DTZ fixtures and an integration test proving a known KRvK probe returns Win, DTZ 23 and 15 moves.
Add FEN.position_key/1 (board, side, castling and en-passant; clocks dropped) and have Tablebase.new/1 key cache reads and writes on it, so transpositions and clock-only differences share an entry. The Lichess category is a position property, so dropping the clocks is safe.
Update the cache docs and tests, and give the play fixture's after_first/after_losing distinct positions since they previously differed only by the halfmove clock.
Add a Backend behaviour, a Lichess implementation, and a Router that picks a backend by piece count: five or fewer to the local slot, six or seven to the remote slot, and eight or more short-circuited as not covered with no network call. Both slots point at Lichess for now; the local Syzygy backend replaces the local slot when it lands.
Map :not_covered to the existing 'not covered by the tablebase' message.
Drop dtm/dtc from Tablebase and Tablebase.Move; the local Syzygy backend cannot produce them and nothing reads them. The Lichess parser now maps only category, checkmate and moves (with per-move category, dtz, san and uci), and the struct-shape tests follow.
Add FEN.piece_count/1 and reject positions with more than 7 pieces in the Endgame changeset, so an 8+ piece FEN fails at entry instead of saving and then blocking at play.
Accumulate all FEN issues rather than letting a later check overwrite an earlier one, and guard the seed FENs with a test (dropping the out-of-range seed that bypassed the changeset).
* Slim the cached tablebase value to the fields the trainer reads (#20)
* Sample for LFU eviction instead of scanning the cache (#21)
* Make cache size configurable and back up daily off-peak (#22)
* Document the cache design in the glossary and an ADR (#23)
* Address cache-scaling code review findings
The admin list now groups endgames under collapsible Key headers and can be
filtered by Theme (search + tappable chips) and Key. Group counts and
filtering run in SQL, and a group's endgames are fetched only when it is
expanded, so the list stays fast as the dataset grows.
- Add Endgames.list_keys/0, list_key_groups/1 and list_endgames_for_key/2.
- Add the shared ChesstrainerWeb.EndgameList filter bar and grouped list
component (to be reused by the player picker).
- Rework EndgameLive.Index around server-side group state.
Closes#15
The Endgame key is now a canonical, colour-agnostic material signature derived
from the FEN instead of a hand-typed field: each side's pieces, strongest side
first by material value (e.g. "KQ v KR"). This fixes the seed ordering
inconsistency and guarantees the key always matches the position.
- Add Chesstrainer.Key.from_fen/1.
- Compute the key in the Endgame changeset; drop the manual input and its
regex validation.
- Index endgames.key and derive keys in the seeds.
- Document the Key term and the decision (ADR 0003).
Closes#13
Rename the endgame-label concept from Tag to Theme to match the glossary:
the context becomes Chesstrainer.Themes, the schema Chesstrainer.Themes.Theme,
and the Endgame association is :themes. The category enum
(endgame/opening/tactic) is preserved.
Rewrite the original migrations in place so themes / endgames_themes
(with theme_id) are created directly, instead of adding a transition
migration.
Closes#12
#11: endgames can carry multiple themes (endgame tags) chosen from a
searchable, autocompleting picker with inline creation. Tags are staged
in the form and saved atomically, so they work on both New and Edit.
Tag names are trimmed and case-insensitively unique per category. Admin
Show and Index display themes, and the seeds ship a starter vocabulary.
The theme picker has its own form: LiveView requires phx-change inputs to
live inside a form, and a separate theme form keeps Enter from
submitting the endgame form.
#7: prefetch the tablebase at mount and grade a played move locally
against its moves list (optimal / winning / losing), blocking the
attempt when the tablebase is rate-limited or unavailable. The client
now reports prev_fen and promotion, and the trainer board locks after a
graded move until reset.
Fix the 429 clause ordering in the tablebase parser so the cooldown
actually fires, and use the HTTPS endpoint.
Add dialyxir to the dev/test deps.
#5: normalise the full lichess tablebase category vocabulary (cursed-win
and blessed-loss classify as draws), return {:ok, tablebase} on a cache
hit, and unwrap the tablebase in game_from_fen.
#6: replace the LRU cache with an LFU cache storing
{key, value, frequency, last_access}, evicting the least-frequently-used
entry (LRU tie-break) and halving every frequency on a weekly sweep.
Persistence moves to a versioned v2 dets file, dropping the legacy
backup; the cache restores synchronously at startup.
Also updates stale tags enum and Play LiveView tests, and drops the
tracked legacy cache backup.
* create endgame play section
* add cache tests
* normalise FENs - trim and fuill move to 1 on addition
* remove svg on root and allow navbar to exist across all pages
* update nav items