Week 12 Studio: Finishing, not adding

The last studio before the final project is due — cutting scope, proving the extension actually works, and drafting the retrospective

10 May 2027

Before the session#

Bring whichever extension from last week’s menu you’ve chosen — pattern- matching completeness, a quickscan-style heuristic, a small collector, or profiling-driven performance work — in whatever state it’s actually in. This is the last studio before the final project is due; the point of today is finishing what you have, not starting something new.

In the session#

• Name what you’re cutting, out loud, before you cut it. Look honestly at how much of your chosen extension is actually going to be done by the deadline, and if the answer is “not all of it,” decide right now which piece you’re dropping rather than discovering it by running out of time silently. A smaller extension that’s fully working and honestly scoped in the retrospective beats a larger one left in an undocumented half-state.
• Get one real test case passing end-to-end. Whatever your extension is, make sure at least one concrete program exercises it correctly before you spend more time broadening it — a BREAKX implementation that hasn’t been run against a subject string yet, or a collector that’s never actually triggered a collection, isn’t evidence of anything yet.
• Re-run the differential-testing harness. Confirm your extended interpreter still agrees with CSNOBOL4 on everything it agreed on before your extension existed — an extension that silently changes behaviour on programs it wasn’t meant to touch is a regression, not a feature, and week 11’s harness is exactly the tool for catching that without checking every case by hand.
• Draft the retrospective now, not the night before it’s due. Write down, in a few honest paragraphs: what you chose to extend and why; what it actually cost, in the time or complexity sense, compared to what you expected going in; and what specifically you understand now about SNOBOL4, or about implementing a language generally, that you didn’t twelve weeks ago. A retrospective written in the studio, while the work is still fresh, is reliably more specific than one written from memory later.

Afterwards#

This is the last studio. What’s left is finishing the extension, the differential-testing pass, and the retrospective, and submitting all of it as the final project by the end of the week. The interpreter you’re submitting has been built, checked, and extended across twelve weeks of sessions — this is where that accumulated work becomes one finished artefact.