ELS 2026
Basics
| Field | Value |
|---|---|
| Dates | – |
| Venue | 2nd floor, Długa 72, 31-146 Kraków, Poland |
| Host | European Lisp Symposium (19th) |
| URL | european-lisp-symposium.org/2026 |
| ICS | programme.ics |
| Format | Single-track, two days, in-person |
| Framing | Lisp practice, MOP, language implementation, codegen, hardware in CL |
Program
Times in CEST (UTC+2). Asterisks mark keynotes.
Day 1 — Mon 2026-05-11
- 09:00 Registration
- 09:45 Welcome and announcements
- 10:00 Sponsored — Jakob Østergaard (Keepit): Choosing Lisp / Commercial Use at Keepit
- Session 1 — Efficiency & Complexity
- 11:30 Max-Gerd Retzlaff — Efficient Range Estimation with NDB Interpreted Code
- 12:00 Jérôme E. Onwunalu — Taming Complexity: Building & Deploying a 270 KLOC Integrated Scientific Application in Common Lisp
- 14:00 Keynote — François-René Rideau: Lambda, the Ultimate Paradigm
- Session 2 — Language
- 15:30 Jan Moringen, Robert Strandh — A Highly Configurable Common Lisp Reader
- 16:00 Jesse Bouwman — Epsilon: A Module System for Lisp
- 16:30 Didier Verna — An Update on the Method Combinations MOP
- 17:15 Lightning Talks
- 18:30 Banquet
Day 2 — Tue 2026-05-12
- 09:00 Registration
- 09:45 Announcements
- 10:00 Keynote — Daniel Kochmański: McCLIM
- Session 3 — Paradigms & Semantics
- 11:30 Frank Adrian — FOL: Bridging Object-Oriented and Functional Programming via the Metaobject Protocol
- 12:00 Daniel Gößwein, Michael Hafner, Kai Selgrad — Ensuring Consistency with OptiX-Semantics: A Use-Case for Low-Effort Code Generation
- 14:00 Keynote — Daniel Kochmański: ECL
- Session 4 — Applications
- 15:30 Stefan-Tiberiu Petre — A Hardware Description Language and Simulation Framework in Common Lisp
- 16:00 Max-Gerd Retzlaff — A Lisp Dialect for NDB Interpreted Code
- 16:30 Jacob 'Screwlisp' Pouw-Waas — New Deep Learning Receiver Operating Characteristic Formulation Made with Lisp
- 17:15 Lightning Talks
Why
ELS sits at the intersection of three live concerns: the MOP as governance primitive, formal semantics in code generation, and Lisp implementations as the actual contract surface beneath everything that calls itself Lisp.
Verna and Adrian are direct MOP probes — the same metaobject
extension points that emacs-mcp-maximalist exposes as MCP tools.
OptiX-Semantics formalizes the consistency invariant that
elenctic-spec leaves at the methodology layer. Petre's HDL grounds
Lisp in hardware-formal territory, a vocabulary sjg recognizes.
Kochmański's two keynotes (ECL, McCLIM) own the substrate layer for
half the program — embeddable CL and the CLIM revival are not the
SBCL-default surface most of the agentic-Lisp discourse assumes.
The Polish CL community runs a different surface than what is visible from a US/SBCL vantage. ECL embedding, McCLIM, NDB interpretation, and the Strandh reader lineage are the actual centers of gravity here. Conference reads partly as recalibration: which parts of the CL stack still carry research weight, which are kept alive by a single maintainer, and which are the new entry points (Epsilon module system, OptiX codegen).
Refutation
If the trip yields no updated priors on any of
MOP-as-extension-point · codegen-as-semantic-refinement ·
reader-as-protocol-surface · ECL-as-embedding-substrate ·
range-estimation-as-abstract-interpretation — the mapping below was
wrong about which terms carry weight in this specific room.
Post-mortem in els-2026-debrief.org.
Speaker × concern mapping
Seven Concerns layers L1 (Substrate) through L7 (Coordination). Mapping is forecast, not observed — revision expected after the proceedings and recordings drop.
| Speaker | Work | Concern | Claim under test |
|---|---|---|---|
| François-René Rideau | Lambda paradigm (keynote) | L3, L7 | Lambda is the paradigm primitive, not a calculus subset |
| Daniel Kochmański | ECL (keynote) | L1, L2 | Implementation choice is the contract surface |
| Daniel Kochmański | McCLIM (keynote) | L6, L7 | MOP-extensible UI as a coordination layer, not a widget kit |
| Didier Verna | Method Combinations MOP | L3 | MOP refinement is type-level extension, not just dispatch |
| Frank Adrian | FOL via MOP | L3, L4 | OO/FP duality dissolves at the metaobject layer |
| Jan Moringen, Robert Strandh | Configurable CL Reader | L2 | Reader is a parser-as-protocol surface, not a constant |
| Jesse Bouwman | Epsilon module system | L4 | Modules carry the provenance partition |
| Jérôme E. Onwunalu | 270 KLOC production CL | L5, L6 | CL scales when the MOP carries invariants out of band |
| Gößwein/Hafner/Selgrad | OptiX-Semantics codegen | L3 | Codegen consistency is semantic refinement, not testing |
| Stefan-Tiberiu Petre | HDL & simulation in CL | L1, L3 | Hardware description as Lisp-formal substrate |
| Max-Gerd Retzlaff | NDB range estimation | L2 | Range estimation = abstract interpretation, CL surface |
| Max-Gerd Retzlaff | NDB Lisp dialect | L2, L3 | DSL design as quotient over interpreter implementation |
| Jacob Pouw-Waas | Deep Learning ROC in Lisp | L4 | Lisp as expression layer for ML primitives, not just glue |
Companion artifacts
els-2026-debrief.org— post-attendance review, mapping revision, lightning-talk capture.mop-as-mcp-tool-exposure.org— Verna + Adrian againstemacs-mcp-maximalistobarray exposure. The MOP IS the tool-list contract.optix-semantics-vs-elenctic-spec.org— codegen consistency as L3 refinement, methodology versus type-level enforcement.hdl-in-cl-formal-substrate.org— Petre's HDL against the formal-methods/hardware vocabulary; sjg co-review.
Aporia
- Single-speaker concentration. Kochmański holds both ECL and McCLIM keynotes. If his framing dominates Day 2, the takeaways collapse to one maintainer's view of the CL landscape rather than the community's.
- Sponsored-talk asymmetry. The Keepit talk is the only production CL signal on the program. Either the most concrete data point of the conference or a recruiting pitch with a Lisp-flavored preamble. Framing collapses one way or the other in the first ten minutes.
- Hybrid attendance, asynchronous review. The Aidan agent
(
computeruse.dron minibos) tracks the program calendar; primary review happens after the fact from notes and proceedings. The hallway-track signal does not survive this transmission. Treat the writeup as a paper-trail forecast revision, not a conference report. - NDB is a Retzlaff-owned domain. Two slots on the program means either the work carries airtime on merit or the selection committee accepted under-subscribed slots. Abstracts are bare; verdict requires the talks themselves.
- Lightning Talks are unpredictable. Standard tradeoff: high surface-area diversity at low formal-preparation cost. Could be the highest-signal blocks of the conference or filler.
- "European" framing is partly geographic, partly cultural. The CL surface visible from Kraków is not the surface visible from Pittsburgh, Tokyo, or Boston. The community knows this and curates for it; the asymmetry is the data, not the noise.
Related
- Scheme — R7RS, hygiene, continuations; the minimalist Lisp tradition that puts ECL/McCLIM in context.
- Interlisp — Interlisp-D / Medley archaeology, prior art for Lisp environments as a coordination layer.
- Agentic Systems 2026 — Seven Concerns framework, MOP as governance-tuple primitive.
- Elenctic Vibe Code Review — Socratic refinement, foil to OptiX semantic consistency.
- Lean4 on FreeBSD — theorem proving toolchain, parallel formal-methods substrate.
- TLA+ System Design — formal specification practice.
- BugBash 2026 — the immediately prior correctness conference; ELS reads as the Lisp-shaped complement.