Compiler Construction — Exam Prep (`2026-winter`)

Course: compiler-construction · Concept map: _index.md · Lecturer: Janis Voigtländer.

Written exam: 11 March 2027 (announced on slide 5 of the 28.10.2026 deck).
Prüfungsvorleistung: four milestones, at least the first three must be reached — (1) lexer for C₀ ~week 4, (2) parser for C₀ ~week 8, (3) semantic analysis for C₀ ~week 10–11, (4) intermediate code generation. Copying implementations (including from LLMs) ⇒ disqualification.
Course language is German, slides are English.

Coverage in the wiki

Block	Slides	In the wiki?	Page
Organisation, compiler phases, C₀, target AM₀	1–18	summary only	course page
Lexical analysis — formalism	19–51	full	lexical-analysis-formalism
Lexical analysis — practice	52–55	full	practical-lexers
Syntactical analysis (CFG, LL(1), AST, SLR)	—	not yet in the deck	—
Semantical analysis (attribute grammars), interpreters, code generation	—	not yet in the deck	—

Weak concepts

Concept	Why weak	Next action
DFA minimization (partition refinement)	slide 49 shows only the result, the algorithm came from literature	re-derive the §6.3 example on paper without looking
IA-32 holes / why x86 needed VT-x	announced concept, not yet in the deck — will reappear under CPU-adjacent material only if the course covers the AM₀ machine in depth	skip for now
Thompson construction state count	the lecture’s optimised variant vs. textbook `≤ 2·	r

Study plan (lexical-analysis block)

Session	Goal	Self-check
1	slide-35 exercises 1–4 (regex semantics, abbreviations, algebra, number words)	can I derive `ac ∈ L((a
2	build NFAs for `a*(a	b)aa`and`((a
3	powerset construction on both NFAs (task 2) + the exercise NFA (task 3)	is every DFA state a closed subset?
4	minimize the two exercise DFAs (task 4)	did I complete with a dead state first?
5	quiz from the wiki, then a timed 45-minute mock on the whole block	wrong answers → `mastery: learning`, re-test next session

Quiz history

Date	Source	Score	Missed concepts
—	—	—	no quiz taken yet

Open questions for the lecturer / exercise session

Is the exam’s automaton notation expected in the lecture’s letters ( $S, s_{0}, F, σ$ / $T$ ), or is Mogensen’s ( $Q, q_{0}, F, δ$ ) accepted?
Are the optimised Thompson variants (as on slide 42) acceptable answers, or is the literal textbook construction required?
Does the milestone lexer have to be generator-based (Alex), or is a hand-written scanner fine? (slide 14 promises “guided hands-on”, so this is worth asking early)