42nd EuroForth 2026

Programme

Forth standard meeting

Times are converted to your local timezone indicated by a blue background, if you have javascript enabled (no, this cannot be done server-side).
Otherwise UTC(+0) will be displayed and you have to calculate the offset yourself.

Wednesday, 4th September 2026 UTC(+0)

• 12:30 Session 1
• 14:30 Coffee Break
• 15:00 Session 2
• 17:00 End of main sessions
• 19:00 Base Camp Indian Kitchen
• Workshops

Thursday, 5th September 2026 UTC(+0)

• 07:30 Session 3
• 09:00 Coffee Break
• 09:15 Session 4
• 11:00 Lunch
• 12:30 Session 5
• 14:30 Coffee Break
• 14:45 Session 6
• 17:00 End of main sessions
• 19:00 Dinner at Giovanni's On The Hayes
• Workshops

Friday, 6th September 2026 UTC(+0)

• 07:30 Session 7
• 09:00 Coffee Break
• 09:15 Session 8
• 11:00 Lunch

EuroForth 2026 conference

on air ... these session are streamed live on twitch, recorded and shared on youtube for future generations to enjoy and to rewatch

Friday, 4th September 2026 UTC(+0)

• 13:00 Session 1 on air

  • Francois Laagel: Implementing Exceptions in Z79Forth/A (30min)
    This talk will describe the path I went through in order to implement exception support in Z79Forth. The Z79Forth platform is an electronic kit that is based on a single board computer centered on the Hitachi HD6309. In essence, it is an 8 bit computing revival effort I initiated back in 2019. Originally targeting 79-STANDARD, it has been adapted for conformance to the ANS94 specification. It has been commercially available for about three years now.

    [TAGS: retro-computing, bare metal programming, embedded systems]

  • M. Anton Ertl: Gforth 1.0 (30min)
    Gforth 1.0 is the first feature release in the 18 years since Gforth 0.7.0. It offers many new features and other changes. This presentation gives an overview about what has changed.

  • M. Anton Ertl: Control Structure Enhancements (15min)
    Generalized Eaker's CASE supports arbitrary conditions and two ways to construct loops. MEM-DO and MEM+DO are convenient ways to walk across arrays.

• 14:15 BioBreak
• 14:35 Session 2 on air
  • Bill Stoddart: Monte Carlo Simulation and False Positives in statistical tests (45min)
    If a certain disease has an assumed incidence of 1% in a population and we use a test that is 99% accurate, a positive test result only suggests you have a 50% chance of having the disease. This is beacause the probability of a false positive is relatively high for relatively rare conditions. Failure to appreciate this has had serious consequences for the justice system. We use reversible Forth to perform Monty Carlo simulation of sample testing with a view to illustrating this phenomena.
  • Axel Prütz Kanne: Towards Reversible Stack-Based Computing (45min)
    Reversible computing has the potential for creating computing systems with very low energy usage, by going under Landauer's Limit using adiabatic circuit design. To fully realize the benefits of reversible computing systems, it requires that all layers of the computer are designed to be reversible, both the hardware and software. This presentation aims to contribute to this field of research, by exploring using stack-based computing for reversible computing. I will present both the design of an abstract reversible stack machine, which is both cleanly reversible and aims to be minimal in instruction set size and required hardware resources, and a reversible stack-based language inspired by Forth.
• 16:05 Workshops / Important night talks
• 18:00 Dinner at Le Monde

Saturday, 5th September 2026 UTC(+0)

• 08:00 Session 3 on air
  • Jaanus Pöial: Configurable Type Checking for Forth (45min)
    Forth words exchange operands through a data stack, and Forth source may also change
    the parser and dictionary while it is being read. This paper describes a configurable source-
    level checker for that setting. The checker represents a phrase by the stack cells it requires
    and the cells it leaves. Each cell has a profile-defined type and may have a symbolic identifier
    that records value flow through words such as DUP, SWAP, and OVER.
    Straight-line code is checked by matching the output of one word with the input of the
    next. Branches are combined only when the implementation can construct one common stack
    interface. Repetition uses a finite test that compares one pass with two passes. The paper
    presents each algorithm through source examples and pseudocode. The examples include
    unequal branch depths and a loop counterexample. The paper also distinguishes implemented
    checks from stronger guarantees. Normalization can discard value-flow information. Moreover,
    the current branch operator is not a sound summary for every program, and the loop test
    does not compute an inductive invariant.
    The type hierarchy, primitive stack effects, parser words, defining words, literals, and con-
    trol structures are supplied in text files. The native implementation is gforth-evaluator.fs.
    A complete artifact manifest and reproducible example are included. Comparisons with
    StrongForth, pluggable Forth typing, and WebAssembly show which restrictions belong to
    this prototype rather than to stack languages in general.
  • Nick Nelson: Does AI speak Forth (30min)
    Surprisingly, the answer is an emphatic yes
• 09:15 BioBreak
• 09:35 Session 4 on air
  • Nick Nelson: Autotranslation within Forth (30min)
    Now, we use AI from within our application. We examine two approaches.
  • Bill Stoddart: Simpson's Paradox illustrated in Reversible Forth. (30min)
    Simpsons Paradox concerns statistical results which occur in multiple groups but reverse when the groups are combined. We use the backtracking abilities of Reversible Forth to search for examples of this strange phenomena.
• 10:35 Lunch
• 12:00 Leave for Excursion
• 12:45 Guided tour of the Welsh Parliament (Senadd)
• 17:45 Dinner at Assador 44
• Workshops

Sunday, 6th September 2026 UTC(+0)

• 08:00 Session 5 on air
  • M. Anton Ertl: Floating-point output (15min)
    E. outputs an FP number in a way that is recognized as FP number on input and produces exactly the same FP number. E.P limits the mantissa length. E.EXACT outputs the exact decimal representation of the FP number.
  • Nick Nelson: Enhancements to Forth Enums (15min)
    We find some useful ways to use the chains and modifiers.
  • Stephen Pelc: Maybe (30min)
    TBD
• 09:00 BioBreak
• 09:15 Session 6 on air
  • Nick Nelson: SNIPPET and INSERT (15min)
    A nice way to place macros within code.
  • Gerald Wodni: VFX6 (45min)
    TBD
• 10:15 End of EuroForth Conference