Skip to content

curriculum(APP18): competency-derived JavaScript language and async evidence - #165

Merged
christophersesugh merged 1 commit into
mainfrom
curriculum/app18-competency-first-reassessment
Nov 18, 2026
Merged

christophersesugh merged 1 commit into
mainfrom
curriculum/app18-competency-first-reassessment

Conversation

@christophersesugh

Copy link
Copy Markdown
Collaborator

M3 / APP18 — competency-first academic reassessment

1. Baseline, authority and scope

2. Independently derived exit competence and disciplinary evidence

Exit capability: An SEF graduate with programming and mathematical/algorithmic preparation can predict actual JavaScript language behavior, implement cohesive functions and modules around explicit data invariants, inspect object and scope failures, enforce safe runtime data contracts, reason about promise scheduling and error propagation, reproduce a maintained local toolchain, test adverse cases, and independently defend one modifiable JavaScript utility. Learners must be able to diagnose an unfamiliar bounded change rather than follow a tutorial.

Boundary and ownership:

  • SEF2/SEF4/SEF10 prepare general programming, algorithms, testing and professional release reasoning: APP18 applies these to JavaScript-specific coercion, numerical/Unicode semantics, object/closure behavior, modules and promises rather than re-teaching generic CS principles.
  • APP18 owns language contracts, object/prototype behavior, Promise composition, local ESM tooling and proportionate tests. APP19 owns TypeScript static inference/structural types and runtime-erasure boundaries; APP21 deepens Node/V8/libuv/OS, streams, buffers, HTTP, processes, service lifecycle. APP2/APP20 own browser application/React work. No HTML/CSS, DOM, cloud, backend or TypeScript elective becomes a hidden prerequisite.
  • Scope is not determined by a preferred number of modules, lesson entries or assessments.

Research benchmarks (sources, not a template or accreditation):

  1. ECMA-262 17th edition / ECMAScript 2026 — language value conversion, lexical bindings, prototype lookup, modules, Promise jobs and semantics; distinguish specification from implementation.
  2. Node.js ESM and package resolution — relative specifiers, package exports, host-dependent module resolution and CJS interop.
  3. Node.js process.nextTick and queueMicrotask documentation — event order can differ between Node ESM and CommonJS; do not claim universal Node/browser event-loop timing.
  4. Node.js built-in test runner — free local behavior tests and failure reproduction without a third-party framework.
  5. Node.js release/support policy — pin a supported runtime baseline for reproducibility rather than teaching a perpetually fixed release number.
  6. MDN Promise.all — rejection of aggregate does not cancel other started operations, a counterexample required by this review.
  7. MIT 6.102 Software Construction, Spring 2026 — observable specifications, invariants, modularity and testing as transferable competence, not JavaScript-specific course packaging.
  8. Existing OEE APP18, APP19, APP21, SEF2 and APP program catalog specify actual neighboring skills and prerequisites.

3. Outcome dispositions and semantic continuity

Retain APP18-O1–O11, unchanged IDs and published meanings. Their decisions remain: O1 language vs host; O2 values/coercion/precision/Unicode; O3 control flow/decomposition; O4 lexical scope/closures/this/aliasing; O5 objects/prototypes/collections; O6 runtime contracts/explicit failure; O7 cohesive ESM APIs; O8 asynchronous work and scheduling; O9 supported local Node toolchain; O10 tests/debugging/cost; O11 independent integrated delivery/change/defense. No indispensable outcome was removed to reduce source length.

Clarify rather than replace: O5's prototype-data observation and O6's untrusted-boundary enforcement are distinct, not two identical JSON/data-parsing classes. O8's Promise-failure criterion is strengthened to demonstrate continuing in-flight work after Promise.all rejection and nonautomatic cancellation; this makes the existing outcome assessable rather than inventing a new language capability. O9 remains small local project reproduction, not APP21 system/runtime service specialization. Historical completion decisions must remain version-scoped.

4. Every module's lesson decisions

Current→revised module Entries Academic disposition
M1 → M1 ECMAScript, hosts and first experiments 6 → 4 Merge course history, first runtime, diagnostic and safe-fixture onboarding into actual first local program and host/language prediction; move detailed tool reproducibility to M9. Preserve Python→JavaScript transfer.
M2 → M2 values, equality, precision, text 7 → 6 Remove standalone dates/localization lesson as nonessential to declared exit evidence; consolidate conversion/edge checks while retaining undefined/null distinctions, numeric/BigInt safety, UTF-16/code point/grapheme differences.
M3 → M3 flow, abstraction and functional reasoning 7 → 6 Combine overlapping pure/decomposition and algorithm narrative; keep branching, loop invariants, function forms, HOFs, recursion and functional side-effect boundaries. Do not re-teach SEF's generic algorithm course.
M4 → M4 scope, closures, receivers 7 → 5 Consolidate separate stack/debugging and captured-loop demos into one causal scope investigation; retain lexical initialization/TDZ, closure capture, receiver/this, alias semantics and negative cases.
M5 → M5 objects, prototypes, copies, classes 7 → 5 Merge fragmented property/accessor tutorials; keep own/inherited/descriptor, shallow vs independent copies and class/prototype equivalence. Move unsafe merge of untrusted keys to M7 trust boundary, explicitly distinguishing it from JSON.parse alone.
M6 → M6 collections, iteration, transformation 7 → 6 Combine raw data-shape, JSON and transformation observations where they serve one record-processing task; keep distinct array holes, sorting, Map/Set identity, iterable protocols and bounded parsing choices. Generator construction is illustrative, not a mandatory independent assessed competency.
M7 → M7 runtime data and failures 6 → 5 Unify duplicated validation/safe-boundary fragments; teach allowed-key validation, rejection versus structured partial results, error causes, no eval/unsafe merge and contract-preserving revision.
M8 → M8 ES modules and interfaces 7 → 5 Put package.json, lockfile, install-script risk and project reproduction in M9. Keep ESM live bindings, cycles, APIs, Node resolution/CJS caveats and independent consumer tests as genuinely module-design responsibilities.
M9 → M9 Node tooling and reproduction 6 → 5 Consolidate environment, offline fixtures and clean-start setup into a coherent host/project verification sequence; no mandatory third-party registry access.
M10 → M10 promises and bounded asynchrony 8 → 7 Integrate stack/event-loop preamble; retain chaining, async/await, overlapping versus CPU work, all/allSettled, cooperative cancellation and deterministic failure diagnostics. Add explicit aggregate rejection ≠ automatic cancellation test.
M11 → M11 testing and change reasoning 7 → 6 Combine normal/boundary/property checks without creating an external framework requirement; retain deterministic async tests, debugger/repair, SEF-aligned cost estimates and compatibility decisions.
M12 → M12 one utility and defense 7 → 5 Replace a seven-step restatement of final assessment procedure with authentic project brief, integration of prior checked work, bounded async feature, unfamiliar correction and independent reproduction/defense.

Result: 12 retained conceptual modules; 82 → 65 numbered lesson entries. Module length remains unequal because there is no pedagogical rule requiring equal-sized modules. Do not interpret the reductions as a numerical quality score; no core JavaScript mechanism or required outcome was dropped.

5. Required summative evidence and adversarial cases

All five existing stable assessment IDs are retained with separately decidable evidence; only two contracts require substantive clarification. The same evolving source may be used:

  • APP18-AS1 (O1/O2/O3/O4): unknown-but-fair coercion, precision, Unicode, closure/receiver and alias predictions; two real repairs, before/after tests, accurate host/language reasoning. Check truthiness/zero and captured bindings; do not pass via unexplained output.
  • APP18-AS2 (O2/O3/O5/O6/O7/O10): a tested domain-transforming ESM with safe input shape, duplicates, own/inherited/prototype-like keys, numeric/Unicode limits, deterministic collection operations and explicit error behavior. JSON.parse is not equated to prototype pollution; a later unsafe merge can introduce it.
  • APP18-AS3 (O4/O6/O8/O10): deterministic controlled promises, language-vs-host order, explicit rejection policy and bounded in-flight work. Added a required Promise.all rejects while other already-started jobs continue counterexample and comparison with allSettled. A timeout/abort signal is not inherently cancellation of an arbitrary Promise/underlying operation; no arbitrary sleep timing tests.
  • APP18-AS4 (O1/O7/O9/O10): ESM package and exact supported runtime, built-in local testing, reproducible scripts and a real/seeded module-resolution defect. Built-in-only projects need no meaningless lockfile; where dependency needed, record reviewed fixture and provenance. Offline reproduction and explicit install-script hazards.
  • APP18-AS5 (O1–O11): one bounded cumulative local utility integrating accepted prior results, a new unfamiliar change/counterexample, clean-state reproduction, explicit evidence references and technical defense. Not five unrelated apps or duplicate evidence bundles. Prior AS1–AS4 evidence cannot be skipped simply because AS5 appears polished.

Cross-check: each required outcome has an independently examinable required work/criteria/evidence path. Every assessment explicitly specifies Outcomes, Work, Criteria and Evidence, with negative/failure cases and no hidden hosted infrastructure. Reproducible tests alone are insufficient evidence of conceptual semantics, scope or async understanding.

6. Cross-course, catalog and historical implications

  • Published SEF10 prerequisite and dependent APP19/APP21 and APP9 closure unchanged; this language-foundation course never becomes TypeScript/React/Node-server course.
  • Removed obsolete public prose claiming TypeScript, browser, Node or capstone migrations are still future. APP README already publishes their course definitions. No catalog metadata change required.
  • Course retains v0.2.0, original outcomes/assessment IDs, same draft status and provisional 12–16-week estimate. Workload needs actual learner/assessor timing; shortening lesson entries is not evidence the learner duration should be shortened.
  • No automatic change in old learners' decisions; evidence is interpreted against its original course/source/specification revision. No protocol/SYS/SEF/AI branch modification.
  • Final APP-wide review must still resolve program-entry evidence ambiguity, elective specialization ownership, APP9 integration and cross-course graduation sufficiency.

7. Locator, severity, consequence and disposition

  • Major resolved, M5/M6/M7: overlapping representations and prototype/JSON safety explanations obscured what is about language mechanisms versus untrusted input validation. Distinct responsibilities and observational checks now explicit.
  • Major resolved, M10 and APP18-AS3: early-rejection Promise.all test previously did not independently require evidence remaining work continues; a competent-looking solution could imply automatic cancellation. Required counterexample and correct recovery policy added.
  • Major resolved, M8/M9/M12 and AS4/AS5: module/public packaging and repeated project checklists caused unnecessary duplicated instruction/evidence. One module-design stage, one host-tooling stage and one evolving utility with separately reviewed criteria now explicit.
  • Minor resolved, M1/M2/M3/M4/M11: isolated introductory/history, detached dates/localization, repeated complexity/stack/diagnostic intros and unnecessary tutorial fragment boundaries removed or integrated.
  • Minor resolved, completion revision narrative: stale prose about “eventual” already-published course migrations removed.
  • Retain with justification: 12 distinct conceptual modules remain (including a separate Node tooling and reproducibility stage), and unequal lesson counts are intentional. No artificial module-count change, target lesson range or per-module test/lab quota.
  • Still unverified: learner workload and outcomes in real teaching, independent assessor use, version-specific engine/browser behavior and accessible learning delivery. Do not substitute code inspection for learner testing.

8. Verification, independent challenge and limits

  • Refetched the committed branch source and compared against pinned base. Exactly one file changed, one commit ahead, none behind at review time.
  • Checked 12 sequential module headings, every module's Outcomes and Lessons sections, all lessons numbered correctly, 65 numbered entries, 11 declared outcome IDs, five required assessment IDs, and Outcomes/Work/Criteria/Evidence for every assessment, with all outcomes appearing in required assessment definitions. No static mismatch observed.
  • Examined tests/test_app_catalog.py and tests/test_curriculum_inventory.py for hardcoded APP18 module/lesson count expectations; none found in these tests. This does not replace running the checks.
  • Cross-checked actual APP19/APP21 neighboring course scope and SEF2 preparation; researched authoritative ECMA/Node specifications and software-construction education.
  • Not executed: repository Python structural validators/regression suite, CircleCI, independent learner work, real external browser/Node behavior experiments or full link checking. No CI-pass, certification, tested or verified claim. Require project CI and maintainer pedagogical review before merge.
  • Duration remains an evidence-limited provisional estimate. No conclusion from lesson count.

9. Decision and branch discipline

Ready for maintainer scrutiny / CI, not an audited-verified learner-ready course. Target main directly on this focused branch; do not merge PR #150 and do not merge/delete this branch without maintainer action. Next in established M3 dependency-valid order after merge: APP19 — TypeScript Programming.

…dence and cumulative utility

Signed-off-by: Christopher S. Aondona <christohybrid185@gmail.com>
@christophersesugh
christophersesugh force-pushed the curriculum/app18-competency-first-reassessment branch from ee8e1ee to 3ee4eb2 Compare November 18, 2026 12:57
@christophersesugh
christophersesugh merged commit 9ddddcc into main Nov 18, 2026
7 checks passed
@christophersesugh
christophersesugh deleted the curriculum/app18-competency-first-reassessment branch November 18, 2026 13:00
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant