Software Language Engineers’ Worst Nightmare
summary
Claim: a compact language BabyCobol models real legacy pathologies so tools must handle punchcard layout, inline PERFORM clauses that allow statements between clauses, name‑based overlays like MOVE CORRESPONDING, computed GOTOs and runtime ALTER of control flow. Evidence: a working language and artifacts on the project site and a paper, used as a teaching exercise and benchmark, implementable within a single MSc course or paper, and still purposely simpler than full COBOL despite real‑world scale (COBOL ≈200–220 billion LOC, one system ≈250M LOC).
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:52The World Runs on Legacy
- 2:45I Solved Problem X
- 4:34Lexical Analysis
- 6:57Syntactic Analysis for LOOP
- 8:04Static Semantics: DFG
- 11:50Dynamic Semantics of GO TO
- 12:36Summary
- 13:36Not Covered in BabyCobol
- 15:18Conclusion
related talks
25mCrossover: Towards Compiler-enabled COBOL-C Interoperability
40mAll of this has happened before, and it will all happen again.
43mLiving with and Leveraging GCC
01h 01mThe Worst Programming Language Ever
01h 30mThe Past, Present and Future of Programming Languages
57mWhy Programming Languages Matter
56mThe Past, Present & Future of Programming Languages
49mWhat Counts as a Programming Language?