Draft Standard

Ribose Standard

•
36817(d) (draft 2026-11-01)
UnitsML specifications — Asciiunits Content Language
CalConnect
Ribose Standard

Draft Standard

Warning for Drafts

This document is not a Ribose Standard. It is distributed for review and comment, and is subject to change without notice and may not be referred to as a Standard. Recipients of this draft are invited to submit, with their comments, notification of any relevant patent rights of which they are aware and to provide supporting documentation.

© Ribose Asia Limited 2026

Ribose Asia Limited

Suite 1, 8/F, New Henry House
10 Ice House Street
Central
Hong Kong

copyright@ribose.com
www.ribose.com

All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on the internet or an intranet, without prior written permission. Permission can be requested from the address below.





Foreword

The unit expression is the connective tissue of the series: it is the content of unitsml(...) in legacy dialects, of the unit(...) constructor in pma, and the payload of every dimension decomposition. This part specifies that language as a grammar-only artifact: a PARG grammar compiled to a checksummed artifact, terminals generated from the register, and a corpus frozen as the conformance reference. No runtime code lives beside the grammar; runtimes are separate implementations (Ruby, TypeScript) of the same artifact.


Introduction

When the unit vocabulary lived inside a parser, every register edit meant a parser edit, and consumers drifted. Generating the terminals from the register — and shipping the grammar as a checksummed artifact that Ruby, TypeScript and Rust runtimes all load — turned that drift into a build failure. This part specifies the result.

1  Scope

This document specifies the expression surface; the terminal-generation rule from the register (CalConnect 36814; the registration of multi-term and prefixed symbols; extender rules; and the artifact contract. It does not define the embedding syntax of host languages (asciimath 2 and pma 1 do that) nor any semantics beyond parsing to a formula tree.

2  Normative references

The following documents are referred to in the text in such a way that some or all of their content constitutes requirements of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies.

PARG grammar language and artifact format, (parsanol/docs)

CalConnect 36810, AsciiUnits syntax specification (v1 base standard, frozen 2025-03; historical)

3  Terms and definitions

No terms and definitions are listed in this document.

terminal

a registered string the grammar matches as one token: a unit symbol, prefix symbol, registered multi-term or prefixed symbol, or a structural keyword.

extender

a continuation mark for interrupted unit series; legal only inside an expression.

artifact

the checksummed artifacts/unitsml.json compiled from the grammar; the normative form of this language.

edition

the recorded grammar revision carried by the artifact.

4  Conventions

Requirement language per CalConnect 36812. A single space separates juxtaposed factors; no space occurs inside a terminal; multi-term registered symbols match as whole tokens before sub-term decomposition.

5  Expression grammar

The artifact (artifacts/unitsml.json, checksummed) is normative; this clause orients. The surface: products (kg m); quotients / and // with exponent inversion (m/s^2); powers ^ (integer and specified-fraction numerators); sqrt(...); parentheses; the dimension symbols dim_L, dim_M, dim_I, dim_Θ, dim_N, dim_J, dim_Φ; and the number 1.

EXAMPLE

Expected outcome. kg m/s^2 parses to mass × length × time⁻²; J/(kg K) groups the denominator; sqrt(m^2) resolves to a length root; 1 is the unity expression.

6  Terminal generation from the register

Every unit symbol, prefix symbol and registered multi-term symbol is generated from the UnitsDB 3 release. The grammar never hardcodes vocabulary: a register edit regenerates the terminals.

EXAMPLE

Expected outcome. Adding a unit to the register lands its symbol as a terminal in the regenerated artifact in the same change; a consumer loading a stale artifact against the new register fails checksum verification rather than silently missing the symbol.

7  Registered and prefixed symbols

Registered prefixed symbols (dB, mbar, kgf, cSt, …) are single terminals, resolved deterministically against prefix+unit decompositions; registration order and precedence are part of the generated tables.

EXAMPLE

Expected outcome. dB matches as one token (never d·b); mbar precedes m-prefixed bar in match order; kgf never decomposes to kg·f.

8  Extender rules

Extenders continue interrupted series inside an expression; an expression MAY NOT terminate on an extender (grammar v2 trailing-extender rejection).

EXAMPLE

Expected outcome. m + … inside a continuation series parses; the bare tail m ⟶ at the end of an expression is rejected with a parse error.

9  The artifact contract

  • The artifact carries a checksum set; every consuming runtime verifies checksums before load.

  • The artifact records its grammar edition; edition 2 carries grammar v2 (registered prefixed symbols, trailing-extender rejection).

  • Grammar edits regenerate the artifact AND the corpus in the same change; consumer-runtime CI (dependent-gems against unitsml-ruby) gates the regeneration.

  • The corpus: 143 cases covering every terminal class and every production of Clause 5, frozen byte-for-byte.

10  Conformance

Conformance class parse/render includes this part. The 143-case corpus is replayed by unitsml-ruby, unitsml-ts and asciiunits-ts; the embedding vectors of CalConnect 36819 Annex B cover host-language integration. Additional test-suite examples are given in Annex A.


Annex A
(normative)

Annex A (informative): additional test-suite examples

EXAMPLE 1

Checksum gate. A stale artifact (one edited byte, stale checksums) aborts the load with the typed checksum error in every runtime.

EXAMPLE 2

Trailing-extender rejection. m/s ⟶ fails; m/s ⟶ m parses as a series continuation.

EXAMPLE 3

Longest-match registration. mbar matches as one registered symbol; dB never matches d × B; kgf never decomposes to kg × f.


Bibliography

[1]  parsanol PARG specification and artifact format, (parsanol/docs)

[2]  unitsml/asciiunits, (grammar source, artifacts, corpus); unitsml/unitsml-ruby#80