INSTRUCTOR GUIDE
Observe understanding.
Make feedback usable.
The website supplies practice and artifacts. Instructor judgment and changed examples establish evidence of understanding.
COURSE SPECIFICATION · ETEC T-104
Course specification
CPIT-455 Software Engineering (II), which corresponds to IT 456 Dependable Systems in the revised IT program plan, presented in the sections of the ETEC course specification template, with the iSCARB teaching and assessment framework mapped to each section. Open the course specification.
مقرر هندسة البرمجيات (2) CPIT-455، ويقابله في الخطة المحدثة لبرنامج تقنية المعلومات المقرر IT 456 Dependable Systems، معروضاً وفق أقسام نموذج توصيف المقرر الصادر عن هيئة تقويم التعليم والتدريب، مع إطار iSCARB. الاطلاع على توصيف المقرر.
USE OF GENERATIVE AI · FCIT POLICY
Use of Generative AI
The course follows the FCIT Generative AI Policy Guide for Course Syllabus of the Faculty of Computing and Information Technology and adopts its Option C, Mixed Approach. Each assessment has its own permitted level: the examinations and the in-class check are AI-Free; Part A of Assignments 6–9 is AI-Free; the Python build and the work after commitment are AI-Assisted, as stated on each assignment page; no assessment is AI-Integrated. Students declare any AI use, and use beyond the permitted level may be treated as academic misconduct. The AI level of every assessment.
يتبع المقرر دليل سياسة الذكاء الاصطناعي التوليدي في كلية الحاسبات وتقنية المعلومات، ويطبّق الخيار C (النهج المختلط): لكل تقييم مستوى مسموح. الاختبارات والتحقق في المحاضرة بدون ذكاء اصطناعي، والجزء A في الواجبات 6–9 بدون ذكاء اصطناعي، والبناء في بايثون والعمل بعد التثبيت بمساعدة الذكاء الاصطناعي كما هو مكتوب في صفحة كل واجب. ويصرّح الطالب بأي استخدام، والاستخدام بما يتجاوز المستوى المسموح قد يُعامل مخالفةً أكاديمية. مستوى الذكاء الاصطناعي في كل تقييم.
USE ISCARB IN YOUR TEACHING
A repeatable teaching sequence
- SEE: begin with a specific situation and a decision that requires explanation.
- EXPLAIN: connect each concept to one purposeful visual and one short takeaway. Keep the full source and qualifications in Details.
- TEST: use two brief retrieval checks. Reveal explanations after students reason individually or with a partner.
- DECIDE: use one application stop to build an artifact, inspect evidence and state a boundary. Transfer the reasoning to a new assignment case.
Download the editable iSCARB teaching planning sheet
Developed by Professor Adeeb Noor. Adapt the subject, objectives and case to your course. Use the actual quality of reasoning and later transfer to evaluate learning; slide completion is not mastery.
Design comparison: MIT 6.102 integrates preparation, exercises and feedback. W3C reflow guidance requires preserving access to information and functionality when content is enlarged. Here, long lecture content is paginated instead of compressed into smaller text. This is an implementation direction, not a claim of institutional equivalence or verified WCAG conformance.
PRESENTER MODE
Teaching notes are for the presenter only
Students do not see teaching notes. To teach with them, open a lecture in presenter mode: use the presenter page, or add ?presenter=1 to any lecture link. Press N, or choose Study & tools → Instructor notes, to open the notes for the current slide. Presenter mode stays on in that browser tab; add ?presenter=0 to switch it off before projecting a student view.
Presenter links: Ch 10 · Ch 11 · Ch 12 · Ch 13 · Ch 14 · Ch 15 · Ch 16 · Ch 17 · Ch 20
ASSESSED ASSIGNMENTS
How students receive STRESS
All assignments (1–9): STRESS opens automatically on the assignment page after the student commits Part A. The page reveals it only after a verified, saved commitment, and the final record shows the commit and reveal times. Revealing that evidence needs no code and no extra Blackboard item.
Set up once per assignment
- Create one Blackboard assignment, Assignment N · Chapter X, with the points of that assignment, in the Assignments category.
- Paste the link to the assignment page and the due date. Students submit the final PDF there. That is the only item.
Python build (Assignments 3–9). Each assignment includes a small build the student writes and tests: a fault tree (3), a server-side access rule (4), a reconciliation (5), an evidence-only fit check (6), an adapter with contract checks (7), an idempotent service (8) and an honest feed display (9). Python runs in the student’s browser with a time limit; nothing is uploaded. From Assignment 4 the student’s own tests are mutation-tested: they run against three or four known buggy versions (the classic mistakes of the chapter), and full credit needs every bug caught, as in test-quality grading at leading engineering schools. Each build record also names a micro-viva change request chosen deterministically from the Student ID (for example, “the owners agree a 3-minute limit: change one thing and show which test changes”). In a two-minute micro-viva, ask the student to make that change live: a student who did not write the code cannot. Each task names one primary source (Vesely, Saltzer and Schroeder, Helland, Garlan, Meyer, Maier) to be used in the build plan. The exported record contains the code, the student’s own tests, the course checks and a code fingerprint. The build is graded inside the second criterion (TRACE, TEST, RECOVER, COMPARE, CONTRACT, FAILURE or INTEGRATE); the 5-point total is unchanged and the written budget is 120–180 words. Reference solutions are in your private materials, not in this repository. Passing every check is necessary for full credit, not sufficient: the written record must use what the build showed.
Ownership check. Pair each assignment with a two-minute micro-viva for a few randomly chosen students: picker and question bank.
Evidence of learning. The evidence page computes, locally in the browser, seven measures with confidence intervals and effect sizes. They are: the concept-inventory gain; the peer-instruction gain from exported class votes; the micro-viva agreement rate; build evidence from the final exports (known bugs caught, REFIT choices); the assessment trajectory, including the within-course comparison of Assignments 1–2 with 3–9 and the correlation with the exams; a 12-item perception survey with Cronbach’s α; and a de-identified dataset of consenting students. It also drafts a results paragraph. The statistics are checked against SciPy on synthetic data. For research use, follow the evaluation plan: obtain ethics approval and a consent or opt-out procedure first.
The private STRESS file is not in the repository. Keep it with your course materials. Already committed drafts keep their identities; a student who committed before this change follows the Blackboard steps on reload.
One preparation contract per chapter
Assign the named source selections and publish the calendar deadline in Blackboard. The five objectives, classroom route and named selections are in chapter assessment scope. Announce any additional required source topic before assessment, provide practice and answer questions. Keep optional toolkit enrichment distinct.
Use the first scheduled Check slot in the next discussion for a short retrieval check on the previous preparation: two or three changed examples in about three minutes, then feedback. This replaces a scheduled interaction; it does not create a fourth recurring stop. Announce any grade contribution within the approved assessment plan.
Do not treat source-page clicks, the local practice score or completed fields as proof of learning. The public preparation questions are practice; they are not secure examination items.
Sample individual understanding fairly
Rotate short oral explanations across students over the teaching blocks, using the same criteria: concept accuracy, mechanism, evidence and limits. Allow an equivalent supervised written response where needed. Announce how the observation contributes to any approved participation mark; this release creates no new course-grade category.
| Chapter | Changed-example oral prompt |
|---|---|
| 10 | Two independent servers use the same requirements. Which failure can still affect both, and why? |
| 11 | The measured exposure doubles while failures remain the same. What changes numerically, and what still cannot be concluded about production? |
| 12 | A static analyzer reports no warnings. Which safety claim would still require different evidence? |
| 13 | If an export endpoint is newly added, which part of the previous authorization claim must be checked again? |
| 14 | The backup starts successfully but staff cannot access it. Has the essential service recovered? What observation decides? |
| 15 | A product fits the features but its supported lifetime is too short. Which reuse-planning factor changes your recommendation? |
| 16 | Your adapter passes the valid-input tests. Give an invalid input and explain the required observable result before running it. |
| 17 | Change one event in your lost-response sequence so that a duplicate is expected even with persistent request identities. Explain the new assumption. |
| 20 | A constituent keeps working after leaving the shared dashboard. What does that tell you about operational independence and central authority? |
Mark the mechanism and the evidence
Assignment 1 remains 4 points; the other assignments remain 5 points. FIT includes the short transfer explanation. Source application belongs in the discipline-specific artifact criterion (FIT in Assignment 1). A generic cautious verdict cannot earn full credit without the chapter mechanism. A justified RETAIN can earn the same credit as REVISE or REPLACE.
Use the four stated levels consistently: 1 for correct, connected reasoning; 0.75 for one minor scope or labeling omission; 0.5 for a consequential technical or evidence-link gap; 0 for absent, materially wrong or invented evidence. Each rubric row now spells out what those levels mean for that criterion. The contrasting practice examples show the difference in answer quality without prescribing one verdict.
Check the five-objective preparation and changed examples alongside the artifact. One case samples selected objectives; it is not a complete examination of the chapter. Map the formal assessment to all five objectives before setting the exam. Code Coverage is practised in the coverage step of every build in Assignments 3–9 (course outcome 11).
The practical hour: Python builds in Assignments 3–9
Lab credit is 0. The practical hour is the coded build, the tests, and the recorded run inside Assignments 3–9, completed in the weekly 90-minute practical session. It is marked. It is not optional.
رصيد المعمل صفر. الساعة التطبيقية هي البناء البرمجي والاختبارات والتشغيل المسجَّل داخل الواجبات 3–9، وتُنجز في الحصة التطبيقية الأسبوعية ومدتها 90 دقيقة. وهي مُقيَّمة، وليست اختيارية.
- Assignment 3 (Chapter 12): the fault tree · course learning outcome 4
- Assignment 4 (Chapter 13): the server-side access rule · course learning outcome 5
- Assignment 5 (Chapter 14): the reconciliation · course learning outcome 6
- Assignment 6 (Chapter 15): the evidence-only fit check · course learning outcome 7
- Assignment 7 (Chapter 16): the adapter · course learning outcome 8
- Assignment 8 (Chapter 17): the idempotent service · course learning outcome 9
- Assignment 9 (Chapter 20): the honest display · course learning outcome 10
Each build has student code, at least three of the student’s own tests, the course checks, a mutation test and a coverage step (course outcome 11). Part A cannot be committed without a recorded run, a second run after one test input changes, and the coverage sentence.
Mark the build in the second rubric row. The TESTS RUN row is required and is not replaced by longer writing. A green report alone is insufficient: the written record must use what the build showed. Ask for the micro-viva change named in the build record.
Close the feedback loop
- Return one specific technical correction and one evidence/scope question using the published criterion.
- Ask for a concise correction or justified retention, and a recheck. Use the assignment’s feedback/revision fields; do not request the entire work twice.
- Compare the preserved first prepared export with the revision. Blackboard, not the local export, confirms actual submission.
- Use the next changed example to see whether the correction transfers to a new case.
Preserve completed prior-edition submissions; require no retrospective redo unless you explicitly assign one. Both editions remain accessible with separate local drafts.
Check workload and learning before claiming success
Initial budgets are 35–55 active minutes including preparation, or 45–65 for the two lab chapters. Ask for actual minutes; use the reported median and range to shorten low-value writing or split preparation if the budget is repeatedly exceeded. Do not publish these estimates as measured results.
Use short parallel pre/post questions, a delayed changed-context question, rubric-scored artifacts and sampled individual explanations. Report the actual results and limitations; without a suitable comparison design, improvement does not establish causal superiority to another course or institution.
Pedagogical reference: MIT 6.102, Spring 2025 documents preparation, short checks, implementation, review and iterative submission. This release has no MIT affiliation or equivalence claim.
Teach one source spine; run three short stations
Keep the original mechanism, definition, table or exception visible. Use Full explanation or the precise required source page rather than shrinking a crowded slide. All nine lectures have 20 main slides and five local objectives; those local objectives are not invented national item codes.
Activate each Station button. The student first thinks for about 30 seconds, discusses for about 60, then writes for about 90. These are adjustable teaching defaults, not measured learning-time claims. Use the authored hint before the model response. Review whether the claim, mechanism, evidence and boundary agree—not field length.
The final public classroom constraint is distinct from the assessed STRESS payload. Do not expose the assessed new information before Part A is committed; STRESS reaches students only after Part A is committed: automatically for Assignments 1–2, through Blackboard from Assignment 3. No separate code or release announcement is required. Preserve existing points and deadlines, and do not require resubmission solely for an interface update.
Use the local objective → artifact → existing-field crosswalk in each assignment. Sample explanations orally. Separate content availability, work completeness, self-report and demonstrated capability. Official NCAAA/Jaheziah mapping still requires the approved program matrix.
Use the map as a shared learning agreement
Open the cover and move to slide 2. Explain the relationships between the five concept branches, then distinguish your role and the learner’s work in the three roadmap stages. Show the unchanged five objectives when setting the scope. After each station, use Chapter map to reconnect the current concept to the whole chapter; return through the branch or Slides.
Do not teach the toolkit as another syllabus. The three stations build the same practice card. Exact source selections and the separate assessed artifact remain required; deadlines, raw points, draft identities and the commitment-before-STRESS sequence are unchanged.
Teach through one case thread
After the concept map, introduce the fictional case and keep it open. Each subject slide now identifies its CASE THREAD. Use that line to reconnect the source concept to the case without inventing evidence. The final slide is an explicit stop: close the class, name the required source review and five-objective check, then point to the existing assignment.
The automated duplicate audit rejects repeated adjacent recap content. Chapter 11 removes the redundant CHECK and DECIDE display screens; their unique learning functions are already provided by the three stations and required five-objective practice.