mirror of
https://github.com/mwisnowski/mtg_python_deckbuilder.git
synced 2025-09-22 04:50:46 +02:00
chore:update documentation for new release
This commit is contained in:
parent
ed780d91e9
commit
f07daaeb4a
2 changed files with 11 additions and 69 deletions
21
CHANGELOG.md
21
CHANGELOG.md
|
@ -13,19 +13,13 @@ This format follows Keep a Changelog principles and aims for Semantic Versioning
|
|||
## [Unreleased]
|
||||
|
||||
### Added
|
||||
- Misc land step: dynamic EDHREC keep percentage range (roll between 75%-100%) via `MISC_LAND_EDHREC_KEEP_PERCENT_MIN/MAX` for more variety in utility land pools
|
||||
- Alternatives: initial land support – when requesting alternatives for a land, endpoint now returns land-only suggestions (basics → other basics; non-basics → other non-basics) with heuristic sub-category narrowing on large pools.
|
||||
- Land alternatives now randomize: 12 suggestions sampled each request from a randomly sized window within the top 60–100 ranked land candidates (per-card, no caching) for higher variety.
|
||||
- Misc land debug CSV exports gated behind `MISC_LAND_DEBUG` or diagnostics flag; not produced in normal runs.
|
||||
- (placeholder)
|
||||
|
||||
### Changed
|
||||
- Misc land step now excludes all fetch lands outright (they're handled earlier); reason recorded as `fetch-skip-misc` in diagnostics CSV
|
||||
- Legacy single-value `MISC_LAND_EDHREC_KEEP_PERCENT` retained as fallback if min/max not defined
|
||||
- Documentation: README and compose files updated with misc land tuning env vars (`MISC_LAND_DEBUG`, dynamic EDHREC keep range, theme weighting multipliers)
|
||||
- (placeholder)
|
||||
|
||||
### Fixed
|
||||
- (placeholder) – no current unreleased land alternatives bugs logged
|
||||
- Step 5 card grid scroll flicker at bottom: added overscroll containment and skip virtualization for small (<80 items) grids to prevent upward jump when reaching end
|
||||
- (placeholder)
|
||||
|
||||
## [2.2.10] - 2025-09-11
|
||||
|
||||
|
@ -54,15 +48,6 @@ This format follows Keep a Changelog principles and aims for Semantic Versioning
|
|||
|
||||
## [2.2.8] - 2025-09-10
|
||||
|
||||
### Added
|
||||
- (placeholder)
|
||||
|
||||
### Changed
|
||||
- (placeholder)
|
||||
|
||||
### Fixed
|
||||
- (placeholder)
|
||||
|
||||
## [2.2.7] - 2025-09-10
|
||||
|
||||
### Added
|
||||
|
|
|
@ -1,54 +1,11 @@
|
|||
# MTG Python Deckbuilder ${VERSION}
|
||||
|
||||
## Highlights
|
||||
- Dynamic misc utility land variety: EDHREC keep percentage now randomly rolls between configurable min/max each build (defaults 75%–100%).
|
||||
- Land alternatives overhaul: land-aware suggestions (basics→basics, non-basics→non-basics) plus randomized 12-card window (random slice of top 60–100) for per-request variety.
|
||||
- Cleaner mono-color utility land pools: rainbow/any-color filler and fetch lands excluded after their dedicated phases; explicit allow-list preserves strategic exceptions.
|
||||
- Theme-aware misc land weighting with configurable multipliers (base + per-extra + cap) via new environment overrides.
|
||||
- Production-friendly diagnostics: misc land debug CSVs gated behind `MISC_LAND_DEBUG` or diagnostics flag (off by default).
|
||||
- UI polish & stability: eliminated Step 5 bottom-of-grid scroll flicker (overscroll containment + skip virtualization for small grids <80 items).
|
||||
- Documentation & compose updates: all new tuning variables surfaced in README, compose files, and sample env.
|
||||
### Changed
|
||||
- Web UI: Test Hand uses a default fanned layout on desktop with tightened arc and 40% overlap; outer cards sit lower for a full-arc look
|
||||
- Desktop Test Hand card size set to 280×392; responsive sizes refined at common breakpoints
|
||||
- Theme controls moved from the top banner to the bottom of the left sidebar; sidebar made a flex column with the theme block anchored at the bottom
|
||||
- Mobile banner simplified to show only Menu, title; spacing and gaps tuned to prevent overflow and wrapping
|
||||
|
||||
## Added
|
||||
- Land alternatives: land-only mode with parity filtering (mono-color exclusions, rainbow text heuristics, fetch exclusion, World Tree legality check).
|
||||
- Randomized land alternative selection: 12 suggestions from a random window size inside the top 60–100 ranked candidates (uncached for variety).
|
||||
- Dynamic EDHREC keep range: `MISC_LAND_EDHREC_KEEP_PERCENT_MIN/MAX` (falls back to legacy single `MISC_LAND_EDHREC_KEEP_PERCENT` if min/max unset).
|
||||
- Misc land theme weighting overrides: `MISC_LAND_THEME_MATCH_BASE`, `MISC_LAND_THEME_MATCH_PER_EXTRA`, `MISC_LAND_THEME_MATCH_CAP`.
|
||||
- Debug gating: `MISC_LAND_DEBUG=1` to emit misc land candidate/post-filter CSVs (otherwise only when diagnostics enabled).
|
||||
|
||||
## Changed
|
||||
- Fetch lands fully excluded from misc land (utility) step; they are handled earlier and no longer appear as filler.
|
||||
- Mono-color pass prunes broad rainbow/any-color lands (except allow-list) using expanded text phrase heuristics.
|
||||
- Alternatives endpoint skips caching for land role to preserve per-request randomness; non-land roles retain cache.
|
||||
- Compose / README / .env example updated with new land tuning variables.
|
||||
- Virtualization system now skips small grids (<80 items) to reduce overhead and prevent layout-induced scroll snapping.
|
||||
|
||||
## Fixed
|
||||
- Step 5 scroll flicker / bounce when reaching bottom of short grids (overscroll containment + virtualization threshold).
|
||||
- Random land alternatives previously surfacing excluded or fetch lands—now aligned with misc step filters.
|
||||
|
||||
## Environment Variables (new / updated)
|
||||
| Variable | Purpose | Default |
|
||||
|----------|---------|---------|
|
||||
| MISC_LAND_EDHREC_KEEP_PERCENT_MIN | Lower bound for dynamic EDHREC keep % (0–1) | 0.75 |
|
||||
| MISC_LAND_EDHREC_KEEP_PERCENT_MAX | Upper bound for dynamic EDHREC keep % (0–1) | 1.0 |
|
||||
| MISC_LAND_EDHREC_KEEP_PERCENT | Legacy single fixed keep % (fallback) | 0.80 |
|
||||
| MISC_LAND_DEBUG | Emit misc land debug CSVs | Off |
|
||||
| MISC_LAND_THEME_MATCH_BASE | Base multiplier for first theme match | 1.4 |
|
||||
| MISC_LAND_THEME_MATCH_PER_EXTRA | Increment per additional matching theme | 0.15 |
|
||||
| MISC_LAND_THEME_MATCH_CAP | Cap on total theme multiplier | 2.0 |
|
||||
|
||||
## Upgrade Notes
|
||||
1. No migration steps required; defaults mirror prior behavior but introduce controlled randomness for utility land variety.
|
||||
2. To restore pre-random behavior, set MIN=MAX=1.0 (or rely on legacy `MISC_LAND_EDHREC_KEEP_PERCENT`).
|
||||
3. If deterministic land alternatives are needed for testing, consider temporarily disabling randomness (future flag can be added).
|
||||
4. To analyze utility land selection, enable diagnostics or set `MISC_LAND_DEBUG=1` before running a build; CSVs appear under `logs/` (or diagnostic export path) only when enabled.
|
||||
|
||||
## Testing & Quality
|
||||
- Existing fast test suite passes (include/exclude + summary utilities). Additional targeted tests for randomized window selection can be added in a follow-up if deterministic mode is introduced.
|
||||
- Manual validation: multiple builds confirm varied utility land pools and land alternatives without fetch/rainbow leakage.
|
||||
|
||||
## Future Follow-ups (Optional)
|
||||
- Deterministic toggle for land alternative randomization (e.g., `LAND_ALTS_DETERMINISTIC=1`).
|
||||
- Unit tests focusing on edge-case mono-color filtering and theme weighting bounds.
|
||||
- Potential adaptive virtualization row-height measurement per column for further smoothness (currently fixed estimate works acceptably).
|
||||
### Fixed
|
||||
- Prevented mobile banner overflow by hiding non-essential items and relocating theme controls
|
||||
- Ensured desktop sizing wins over previous inline styles by using global CSS overrides; cards no longer shrink due to flexbox constraints
|
Loading…
Add table
Add a link
Reference in a new issue