Skip to content

Probability Lab improvement roadmap #7

Open
0 of 10 issues completed
Open

Description

@hananather

Notes

Improve Probability Lab as a free learning platform with mathematical depth, usable interactive explanations, accessible practice, and durable learner progress. Preserve working routes, lesson content, interactions, and saved progress while replacing fragile infrastructure incrementally.

A module is ready when independent review scores at least 4/5 in each of correctness, functionality, accessibility, clarity, explanatory depth, and visual/interaction quality, with no blocking defect. Review scores express reviewer judgment; improved learning outcomes require learner evidence. Each changed flow needs meaningful regression tests and hands-on browser checks at desktop and narrow widths, with reduced-motion behavior reviewed.

Baseline revision: a8a58bd. Build and lint pass; no automated assertion suite is configured. Source has 101 page files and seven substantive chapters; Chapter 8 is a placeholder.

Decisions so far

First implementation checkpoint: draft PR #19, revision c6ba8fd. Its Linux/Node24 CI passed257 tests, full JSX lint (0 errors,114 existing warnings), production build, and107 built-page HTTP checks. Runtime npm audit reports0 known findings;5 development-only lint-chain findings remain documented. Local guidance was retained on disk and removed from tracked files.

Implemented repairs include quiz answer/timer/flag persistence and partial review; mobile/keyboard navigation and lesson resume; numerical distribution/paired-test/quiz/calculator corrections; independent calculator references; homepage motion work; verified curriculum counts; and the canonical progress migration contract. This checkpoint is not a claim of whole-platform correctness or deployment.

Foundation review has found and repaired further teaching and interaction defects. The final rebuilt Explorer gate remains in progress; proposed fixes are not awarded a pass before browser recheck. Local durable guest storage is under independent persistence/reset/ownership review before UI integration. Focused learning and spaced review are tracked in #18.

Fog

Long-term content depth and curriculum extensions will be specified after page-level audits. Authentication deployment details depend on the normalized progress contract and the available hosting/database services. Performance targets need reproducible measurements on representative interactions. Content provenance and licensing need review before declaring the whole repository openly licensed.

Activity

  1. hananather commented on Oct 5, 2026

    @hananather
    OwnerAuthor

    Saved checkpoint at 8dc2609 on codex/platform-foundation, in draft PR #19. All current implementation edits are committed and pushed; the source checkout is clean. Work is pausing at the maintainer's request.

    Completed in this wrap-up:

    • Canonical Sample study record and device-private learning-mode resume, preserving visit-only calculations/practice and explicit study completion.
    • Full computed normal-quantile sample-size counts and approximate/ceiling worked steps, including four reproduced rounding defects and valid numerical extremes.
    • Pending account-progress presentation, bounded keyboard focus return after sync, and honest immutable quiz duration labels.
    • Math-rendering contributor guide aligned with the actual hook/wrapper/runtime and text-safety boundary.

    Validation: 1,210 tests passed across 100 executed files, one explicitly skipped provider fixture; ESLint zero errors/109 remaining warnings; production build and all 110 page HTTP checks passed. Exact-commit hosted Quality checks passed. Independent math and presentation reviews passed. Actual browsers confirmed boundary counts, phone bounds, saved modes/study, pending UI, keyboard sync and the retained quiz result. One explicit local-test study fact synchronized to the other browser as 2/66 studied; its reading mode remained private. All seven original quiz attempts/grades and the original applied recovery receipt were unchanged.

    Resume with Sample selected-state accessibility, a readable numerical Explore alternative and mobile equation/control reflow, followed by its learner-perspective acceptance. Then continue the full route/module/question matrix, additional native ownership/offline/reset and cross-version history checks, architecture/motion work, focused learning/spaced review, content depth and rights/licensing. Current inventory: 104 source pages, 110 built routes, 114 canonical activities and 131 bank questions. Most six-dimension ratings remain unreviewed; no whole-platform or production-provider certification is claimed. Issues #8–#18 remain open.

    The canonical Desktop folder retains a local exact resume checkpoint and all private test/review evidence. No production merge or release is included in this checkpoint.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions