CLP-12-01 deep review — September 29, 2026¶
The study guide maps all 29 numbered objectives, grouped 4/3/3/4/4/3/3/5, and expands all 30 original answers. The active July 24, 2025 syllabus retains eight blocks, 55 questions and a 1,000-point normalized maximum with a 70% threshold. The automated objective request timed out; manual comparison retained the snapshot. Review is same-context AI work with human review pending.
Research and contradictions¶
52 sources have explicit reading boundaries. Direct outcomes: {'error': 38, 'ok': 13, 'blocked': 1}. Complete Edube C Advanced public landing, Microsoft ABI/process/index content, selected WG14 clauses, SEI rules, Linux/POSIX references and GMP sections support the teaching. Failed direct requests are not counted as successful reads; bounded public fallback is recorded separately. Nine attempted references had no full body read. No full standard, linked-book, enrolled-course or paid interior audit occurred. Cisco returned a 46-character shell, so an inferred course duration was removed.
The canonical exam page states no formal prerequisite, while indexed issuer badge criteria list both CLA and CLP passes. This conflict remains unresolved. Complete catalog/policy/scheduling bodies from the immediately preceding CLA review were reused within the same context for their explicit CLP scope. Pearson's 15-day failed-retake rule is distinct from entry-level TestNow. Anytime-before/24-hour appointment change wording conflicts; actual booking and individual eligibility were not inspected.
Repairs and original execution¶
The old variadic fragment omitted a required type declaration, and descriptor failure handlers were empty. Five complete original C11 programs now teach promoted variadic contracts, independent va_list copies, bounded formatting, overlap-safe memory operations, extreme-value comparators, locale-state conversion, permitted setjmp context, bounded threaded queues, numerical precision/GMP, descriptor progress/ownership and framed streams. Microsoft x64 ABI register behavior is distinguished from language rules. The guide does not adopt flawed fence-only race repairs or portable-ISO-incompatible setjmp examples from secondary/vendor snippets.
Seventeen workbook build/runs passed on hosted GCC 15.2/Clang 20.1. Both compilers passed 31 language checks, 11 descriptor checks and 69 frame checks. Each queue run processed ten batches of 1,000 items with sum 500,500; each numeric run observed naive=0/compensated=1 and verified exact 2^100 and 3/10 with linked GMP6.3.0. Three optimized variants and four Clang address/undefined sanitizer runs passed. Seven independent header/dialect probes produced four accepted builds/runs and three expected diagnostic rejections; no undefined runtime example was executed.
ThreadSanitizer compiled the queue but crashed with a SEGV. The same failure occurred for a minimal C11 create/join program on both compilers, all exit 66. The cause remains unresolved; there is no clean race-detector result and no runtime/permission change. The queue's separate synchronization argument covers protected predicates, worker-only totals and join before destruction. The resolved WG14 condition-wakeup issue supports predicate rechecking; no spurious wake was forced or claimed.
Real disposable pipe/file/local-stream operations were exercised. EINTR, zero-progress/write-error and thread-start failure are explicit seams, not actual resource exhaustion or delivered signals. Linux close semantics are labelled; no close/read/output failure was induced. The exclusive file is removed on ordinary completion and returned-error cleanup, but a diagnostic CHECK exit may leave its own fixture. Local AF_UNIX framing is not a TCP deployment or packet capture. Eight broader learner labs remain proposed; no Win32 execution, multi-client service, complete cross-platform scenario or MPFR result is claimed.
The operation record preserves exact public source hashes, bounded execution receipts, source-reading boundaries and objective mapping. Full unit/repository/catalog/strict-site gates are attached after completion. The result is reviewed-with-blockers, with source/appointment, runtime-tool, broader-lab and human-review limits retained. No install, persistent service, authentication/permission change, account, purchase, exam attempt or external message occurred.