Release ledger

What's new in SimDeck

A version-by-version record of the improvements, mirrored from the SimDeck release ledger in Linear.

Version 0.58

August 2, 2026

Safer support details and a private student demo summary

Student demo mode now reports only the selected student's cards, support guidance no longer asks for access credentials, and the public homepage remains usable when platform font metrics are wider.

Why it matters

- Students preview only their own active and used card counts instead of seeing whole-class totals.
- Support requests include useful non-sensitive context without copying an access code, password, or personal student information.
- The public homepage wraps safely at 320 CSS pixels across platform font differences.

Verification

- Focused unit and browser regressions cover safe support wording and teacher/student summary separation.
- Cross-browser public smoke retains a strict no-horizontal-overflow assertion at 320 CSS pixels.
- Typecheck, lint, build, unit tests, and exact-head CI remain required before delivery.

Source: Linear release ledger document, PI-213

Keeps Students, Active cards, and Used cards as teacher-only whole-class summary values.
Labels the student summary as Your active cards and Your used cards and updates it immediately after card use.
Requests only class name, approximate time, attempted action, surface, and browser or device type for support.
Makes homepage hero and preview containers intrinsically shrinkable instead of clipping overflow.

Version 0.57

July 31, 2026

Undo recognizes the exact classroom state after an Appwrite reload

Fresh Use, Transfer, manual-add, and role-deal history no longer asks David for a manual correction just because Appwrite spells the same UTC timestamp differently after commit.

Why it matters

- A newly completed classroom action keeps its verified Undo control after the page reloads.
- Real card or role-assignment changes still disable Undo instead of risking an unsafe reversal.
- The reversal rechecks and updates the exact current Appwrite row inside one transaction.
- The homepage product preview now behaves like a complete keyboard tab interface instead of making arrow-key users Tab through every option.

Verification

- All 498 unit tests pass across 51 files, including Appwrite timestamp round trips and undo compare-and-set filters.
- Exact deployment b64554c passed disposable-class Role Deal, Mark Used, and Transfer Undo with production integrity green after cleanup.
- The source-blind public keyboard and 200%-equivalent reflow pass found the homepage arrow-key gap; the focused browser regression now passes locally.
- Typecheck, lint, static security, production build, coverage, and git diff validation pass on the superseding candidate.
- Exact deployment plus a source-blind arrow-key rerun remain required before final delivery.

Source: Linear release ledger document, PI-213

Recognizes JavaScript's Z and Appwrite's +00:00 as equivalent only when the complete UTC timestamp is otherwise identical.
Uses the transaction-read Appwrite timestamp for the final compare-and-set write after equivalence is proven.
Covers single-card actions, every card created by a role deal, and the role-assignment generation.
Adds wrapped arrow-key navigation, Home and End support, roving focus, and explicit tab-to-panel relationships to the homepage preview.

Version 0.56

July 16, 2026

Preview checks now prove the backend state they actually target

SimDeck's browser suite can now distinguish a production-like Appwrite deployment from an intentionally backendless protected preview, so every public test runs without weakening either environment's expectations.

Why it matters

- A healthy backendless preview is no longer mislabeled as broken simply because its hostname ends in vercel.app.
- Production-like checks still require ready Appwrite health and Maya's live demo inventory unless the runner explicitly declares a backendless target.
- The same complete cross-browser suite proves structured degraded health and safe student recovery instead of skipping those cases.

Verification

- All 484 unit tests pass across 49 files, including local, remote-live, explicit-backendless, and invalid-value behavior.
- All 465 public smoke and accessibility cases pass unfiltered across Chromium, Firefox, WebKit, Mobile Chrome, and Mobile Safari against the protected backendless Version 0.55 preview.
- Typecheck, source lint, and git diff validation pass before the exact Version 0.56 staging cycle.
- The Version 0.56 draft remains on PI-213 until the exact commit receives staging proof; the canonical ledger follows that proof.

Source: Linear release ledger document, PI-213

Added an explicit PLAYWRIGHT_EXPECT_LIVE_BACKEND contract with live, backendless, and compatibility-default behavior.
Shared the contract between smoke checks and screenshot readiness so the two QA paths cannot infer different backend states.
Rejected ambiguous flag values instead of silently choosing a weaker assertion path.
Documented exact commands and expectations for production-like and intentionally backendless protected previews.

Version 0.55

July 16, 2026

Student card actions recheck safety and policy at commit time

Student use, direct sends, and approval-required send requests now repeat every safety, participant, and card-policy prerequisite inside the atomic write, so a stale page cannot outrun David's latest classroom controls.

Why it matters

- A simulation lock or disabled student action takes effect before the next card write, even when a student submitted from an older open page.
- Changing send approval mode cannot let an in-flight action take the old direct-send or request path.
- A removed classmate, changed student access, or newly restricted card route stops the action before either the card or Activity history changes.

Verification

- All 480 unit tests pass across 48 files, including stale game safety, approval mode, and template routing decisions.
- Critical-source coverage passes at 85.39% statements, 72.50% branches, 96.17% functions, and 94.45% lines; actions remain at 94.98% line coverage.
- Typecheck, zero-warning source and static-security lint, production build, backendless demo, 10 skeleton states, theme contrast, npm audit with zero vulnerabilities, and git diff validation pass.
- The Version 0.55 draft remains on PI-213 until the exact commit receives staging proof; the canonical ledger follows that proof.

Source: Linear release ledger document, PI-213

Repeated current game safety, acting-student eligibility, optional recipient eligibility, and template policy reads through each student action transaction.
Required approval mode to match the path selected before the transaction, aborting when David changes that setting during submission.
Added game, student owner, and original template identifiers to the exact card compare-and-set guards for use, send, and request reservation.
Covered transaction-time game-lock, approval-mode, and routing-policy drift with focused no-partial-write regressions.

Safer support details and a private student demo summary

Student demo mode now reports only the selected student's cards, support guidance no longer asks for access credentials, and the public homepage remains usable when platform font metrics are wider.

Why it matters

- Students preview only their own active and used card counts instead of seeing whole-class totals.
- Support requests include useful non-sensitive context without copying an access code, password, or personal student information.
- The public homepage wraps safely at 320 CSS pixels across platform font differences.

Verification

- Focused unit and browser regressions cover safe support wording and teacher/student summary separation.
- Cross-browser public smoke retains a strict no-horizontal-overflow assertion at 320 CSS pixels.
- Typecheck, lint, build, unit tests, and exact-head CI remain required before delivery.

Source: Linear release ledger document, PI-213

Keeps Students, Active cards, and Used cards as teacher-only whole-class summary values.

Labels the student summary as Your active cards and Your used cards and updates it immediately after card use.

Requests only class name, approximate time, attempted action, surface, and browser or device type for support.

Makes homepage hero and preview containers intrinsically shrinkable instead of clipping overflow.

Undo recognizes the exact classroom state after an Appwrite reload

Fresh Use, Transfer, manual-add, and role-deal history no longer asks David for a manual correction just because Appwrite spells the same UTC timestamp differently after commit.

Why it matters

- A newly completed classroom action keeps its verified Undo control after the page reloads.
- Real card or role-assignment changes still disable Undo instead of risking an unsafe reversal.
- The reversal rechecks and updates the exact current Appwrite row inside one transaction.
- The homepage product preview now behaves like a complete keyboard tab interface instead of making arrow-key users Tab through every option.

Verification

- All 498 unit tests pass across 51 files, including Appwrite timestamp round trips and undo compare-and-set filters.
- Exact deployment b64554c passed disposable-class Role Deal, Mark Used, and Transfer Undo with production integrity green after cleanup.
- The source-blind public keyboard and 200%-equivalent reflow pass found the homepage arrow-key gap; the focused browser regression now passes locally.
- Typecheck, lint, static security, production build, coverage, and git diff validation pass on the superseding candidate.
- Exact deployment plus a source-blind arrow-key rerun remain required before final delivery.

Source: Linear release ledger document, PI-213

Recognizes JavaScript's Z and Appwrite's +00:00 as equivalent only when the complete UTC timestamp is otherwise identical.

Uses the transaction-read Appwrite timestamp for the final compare-and-set write after equivalence is proven.

Covers single-card actions, every card created by a role deal, and the role-assignment generation.

Adds wrapped arrow-key navigation, Home and End support, roving focus, and explicit tab-to-panel relationships to the homepage preview.

Preview checks now prove the backend state they actually target

Why it matters

- A healthy backendless preview is no longer mislabeled as broken simply because its hostname ends in vercel.app.
- Production-like checks still require ready Appwrite health and Maya's live demo inventory unless the runner explicitly declares a backendless target.
- The same complete cross-browser suite proves structured degraded health and safe student recovery instead of skipping those cases.

Verification

- All 484 unit tests pass across 49 files, including local, remote-live, explicit-backendless, and invalid-value behavior.
- All 465 public smoke and accessibility cases pass unfiltered across Chromium, Firefox, WebKit, Mobile Chrome, and Mobile Safari against the protected backendless Version 0.55 preview.
- Typecheck, source lint, and git diff validation pass before the exact Version 0.56 staging cycle.
- The Version 0.56 draft remains on PI-213 until the exact commit receives staging proof; the canonical ledger follows that proof.

Source: Linear release ledger document, PI-213

Added an explicit PLAYWRIGHT_EXPECT_LIVE_BACKEND contract with live, backendless, and compatibility-default behavior.

Shared the contract between smoke checks and screenshot readiness so the two QA paths cannot infer different backend states.

Rejected ambiguous flag values instead of silently choosing a weaker assertion path.

Documented exact commands and expectations for production-like and intentionally backendless protected previews.

Student card actions recheck safety and policy at commit time

Why it matters

- A simulation lock or disabled student action takes effect before the next card write, even when a student submitted from an older open page.
- Changing send approval mode cannot let an in-flight action take the old direct-send or request path.
- A removed classmate, changed student access, or newly restricted card route stops the action before either the card or Activity history changes.

Verification

- All 480 unit tests pass across 48 files, including stale game safety, approval mode, and template routing decisions.
- Critical-source coverage passes at 85.39% statements, 72.50% branches, 96.17% functions, and 94.45% lines; actions remain at 94.98% line coverage.
- Typecheck, zero-warning source and static-security lint, production build, backendless demo, 10 skeleton states, theme contrast, npm audit with zero vulnerabilities, and git diff validation pass.
- The Version 0.55 draft remains on PI-213 until the exact commit receives staging proof; the canonical ledger follows that proof.

Source: Linear release ledger document, PI-213

Repeated current game safety, acting-student eligibility, optional recipient eligibility, and template policy reads through each student action transaction.

Required approval mode to match the path selected before the transaction, aborting when David changes that setting during submission.

Added game, student owner, and original template identifiers to the exact card compare-and-set guards for use, send, and request reservation.

Covered transaction-time game-lock, approval-mode, and routing-policy drift with focused no-partial-write regressions.