PCET-30-01 deep review — September 30, 2026

The guide maps all 22 numbered objectives and expands all 17 answers. Its three exact original Python files pass 25 test methods and 55 probe checks in each of normal and optimized modes, plus selected lint. Publication does not establish independent human review or completion of the ten broader labs.

Scope and source boundaries

The complete indexed canonical syllabus, including all detailed sub-bullets and the minimum-qualified-candidate profile, was read after direct retrieval and the objective monitor timed out. The September 2 snapshot was retained after manual comparison, not rewritten as a successful automated observation. Its 22 objectives group 4/6/6/6, while 35 exam items group 6/8/10/11 with weights 17.1/22.9/28.6/31.4. The full PDF was not read.

The header, alignment and complete credential page identify active PCET-30-01; introductory syllabus prose says 30-02 and refers to data analysis, while Objective 1.4 has an unrelated storage heading. The guide preserves all testing content and the unresolved publisher conflicts. The page confirms seven-year validity, 35 questions, 45 minutes plus NDA, 75% cumulative threshold, English/Spanish, TestNow, USD 69 exam or USD 86 with retake and a seven-day failed-retake wait. The dedicated practice product remains in development. The PCET-specific policy could not be read; unrelated certification policies were not substituted.

There are 21 direct receipts: 16 successful responses, three errors, one blocked book page and one missing Flake8 candidate. The verified stable Flake8 index is registered separately from the 404 candidate. Source-reading boundaries distinguish complete public pages, selected manual contracts, indexed fallback and unread fetched pages. Generic current unittest/logging pages were retrieved but not read for this review; pinned Python 3.13 passages support the actual examples. PEP 8's fresh body matches the preceding PCAD receipt, allowing explicit reuse of the selected line-length passage. Original fetch timestamps remain September 29 where applicable; review completion is September 30.

Original requirements and execution

Six original teaching contracts cover exact-type/inclusive quantity validation, regular/priority fees, zero/one/many summation, caller-owned JSON streams, validation-before-collaborator receipt behavior and a toy three-failure lockout. Fees are tokens, the client is an in-memory substitute and the lockout is not a real authorization control. Explicit expected values, exception classes, state transitions, restored patches and ownership checks provide requirement-derived oracles. Empty summation intentionally does not validate an unused priority argument; the guide states that limit.

utility.py, test_utility.py and probes.py are embedded exactly and checked against the executed private files. Existing CPython 3.13.14, coverage.py 7.15.4 and Ruff 0.16.4 were used without installing anything. The 25 TestCase methods pass with warnings promoted to errors, normally and under -O; subtest cases are not added to that count. Each probe run passes 55 explicit checks, and normal/optimized JSON reports match. All three files pass selected E4,E7,E9,F,W,E501 rules at 79 columns. Exact sources, hashes, commands, full outputs, timestamps and calculated results are retained in operation evidence.

The test-double examples include unused dummies, a fixed clock stub, a working StringIO substitute, a failure stub, an in-memory client fake, a wrapping spy, an autospec mock and a restored patch where the function is looked up. Autospec checks interface shape, not real provider behavior. Lab 4's erroneous “least suitable” wording is corrected to “least powerful suitable.” Fresh state, specific assertions and cleanup make tests independent; no real API, filesystem integration or acceptance claim follows from fakes.

Observed coverage, debugger and refactor results

The original rebate module has five executable statements and two branch destinations. One True input with a type-only oracle covers 5/5 statements and 1/2 branches, combined 6/7 ≈ 85.71%, missing arc 6→8. False and True cover both branches. Deliberately changing the member value from 80 to 81 still gives 100% coverage and passes the weak oracle. Independently specified expected values [100, 80] detect member True returning 81. The expected mutant mismatch is evidence of detection, not a repository test failure. No production coverage, function-coverage total or general mutation score is claimed.

Two fee representations are checked against a handwritten table for all 20 valid quantity/priority combinations. This exhausts only that finite domain; invalid inputs and arbitrary refactors are outside the comparison. Requirements remain necessary because characterization can preserve a defect.

A real scripted stdlib Pdb session inspects the original off-by-one defect 1 <= quantity < 10. It prints quantity 10, sets a temporary breakpoint at the return, observes accepted False, steps and continues. A print trace independently shows expected True/actual False. Tracing state is restored. The correctly inclusive utility has explicit regression oracles, while the seeded fixture is intentionally retained for diagnosis. No interactive IDE session or deployed repair is claimed.

An isolated assert-based validator rejects zero normally but accepts it under optimization. The actual utility uses explicit exceptions, and TestCase/probe checks remain active in both modes. A named logger records all five levels and restores handler/level/propagation without changing the root logger. An isolated Ruff F401 probe flags an unused import (exit 1) but passes the wrong token value (exit 0), illustrating a particular static rule's limit. Temporary modules/reports remain under the workbook directory and are removed; no coverage database or persistent service is left.

Learning resources and outstanding work

The complete public PT101 outline distinguishes free Core lessons/module tests from USD 49 Full/Pro interactive features, assessment and diploma. Four weeks at about one hour/day and 38 labs are provider claims, not completed course evidence. The Pro assessment does not establish launch of the separate dedicated PCET practice kit. The complete ISTQB overview describes a different exam; linked syllabus/sample questions were not read and its passing rules are not transferred to PCET. The blocked book has no verified runtime/access claim. Other suggested budgets are explicitly author estimates.

Pylint and Flake8 are absent and unexecuted. Selected Pylint development documentation 4.2.0-dev0 and the complete Flake8 7.4.1 index/quickstart were read, but prepared commands and Ruff do not substitute for executing the named tools. Coverage documentation 7.16.2 also differs from the installed version. None of these observations changes an exam's required runtime.

The ten broader labs remain proposed beyond executed components. The plan/classification work, independent inspection, named-tool runs, interactive IDE exercise, broader state exploration and 30-minute exploratory session still need learner work. Human content/accessibility review and readiness approval are pending. Publisher wording, inaccessible policy/PDF and learning-product boundaries remain visible, with no account, purchase, booking, private questions or enrolled material accessed.