PCPP-32-101 deep review — September 29, 2026

The study guide maps 24 numbered objectives, grouped 12/3/3/4/2, and expands all 23 original answers. This same-context AI review leaves human review pending. The complete indexed canonical syllabus was manually compared after direct retrieval and the automated monitor timed out; the existing snapshot was retained, not silently refreshed.

Exam and learning-source findings

The complete exam page still identifies101 as active with lifetime validity and102 as developing with five-year validity and no announced release date. Its single/multiple-select format differs from broader coding/scenario/interactive wording in the syllabus introduction. Variable item points and a normalized 70% score do not establish a fixed count of correct answers out of 45. Only a search-visible PDF table was readable, with section raw maxima totaling 120; the full PDF and current scoring algorithm were not verified.

The complete Pearson policy retains differing anytime-before/24-hour rescheduling passages. Actual appointment terms, eligibility, voucher redemption and OnVUE system tests were not accessed. No account, purchase or external message occurred.

All five public Edube course outlines were read completely: Advanced1, 2, 3, 4, 5. Their provider estimates total 115 hours, correcting the prior combined 80–120-hour planning range. All associate with 32-101 and recommend Essentials 2; the first retains stale coming-soon copy. No enrolled lessons or assessments were read. The complete Pluralsight response is a broad software-development catalog, not a verified PCPP1 path. O'Reilly 403 and unavailable TkDocs prevent interior/duration verification.

There are 44 registered source receipts:36 successful HTTP responses,7 errors and1 blocked response. Per-source reading boundaries distinguish complete pages, selected primary sections, excerpt fallback and unread downloaded aggregates. The generic Python index is 3.14.7 and selected3.13 pages identify 3.13.15; actual execution used the existing 3.13.14 runtime. No full-library, full-W3C-specification or current-typing-specification audit is claimed.

Original execution and teaching repairs

Five exact public files produce four programs: 71 OOP, 34 networking, 37 storage and 31 GUI checks, totaling 173 per mode. All four pass normally and with -O, with ResourceWarning promoted and no stderr. The selected Ruff 0.16.4 rules E4,E7,E9,F,W,E501 pass at 79 characters. File hashes, exact commands, timestamps and outputs are preserved; these are original fixtures, not copied course labs or certification questions.

OOP examples cover cooperative diamond MRO and composition, built-in subclass copy/mutation behavior, decorator evaluation/application/call order, method types, ABC virtual registration, properties, exception cause/context, metaclass creation, shared/cyclic copying and trusted pickle/shelve persistence. The observed shelf backend is dbm.sqlite3. A failed property update preserves prior state, and optimized runs retain all checks.

The networking program uses a disposable IPv4 loopback HTTP service for GET/POST/PUT/DELETE, 204 handling, status/media/schema/size failures, malformed JSON, redirects and read timeout. A separate real TCP fixture checks bounded length framing, multiple frames, short reads, early EOF, oversize and idle timeout. No packet-boundary assumption, total-timeout guarantee, TLS or production authentication test is made.

Storage uses explicit Python 3.12+ autocommit=False, parameter binding and separate connection ownership. Tests verify rollback after a delete and partial failed batch, committed update/delete, uncommitted-close rollback, quoted multiline CSV, XML schema/namespace behavior, logging thresholds and ConfigParser conversion/interpolation/fallback errors. A tiny original DTD violation parses with ElementTree, distinguishing tree parsing from validity checking. All XML/pickle/shelf input is trusted and created in the run.

The withdrawn Tk desk integrates abstract HTTP/file sources, decorated tracing, a worker queue, event-loop polling, SQLite snapshots, configuration, CSV/XML and logging. It tests button/bound-event/variable/radio callbacks, Canvas items, input validation, captured dialogs, retained state after source errors, responsiveness during a gated read, callback cancellation, observer removal and root destruction. The old blanket geometry-manager ban is corrected: pack/grid conflict in one container, while grid/place can coexist for different children. No visible dialog, manual appearance/accessibility assessment or full proposed history/retry application was completed.

Temporary directories were checked under the dedicated workbook folder and removed; sockets, workers, handlers, database connections and the Tk root were closed. Twelve broader learner activities remain proposed. The reviewed-with-blockers outcome retains access, format/scoring/policy, full-lab and human-review limits. No installation, persistent service, reboot or runtime/authentication/permission change occurred.

The operation record contains exact source, execution receipts, 24 objective mappings and all reading boundaries. Full unit/repository/catalog/strict-site gates are attached after they complete.