Feature layer identifiers: from advice to rule (universal)

Gist. Layered features such as Number[psor] have a layer identifier. Lowercase Latin letters in it used to be only recommended; now they are required. The trigger seems to have been Old Georgian: after 2.18 its layers sauf2 and stack2 were renamed to dsauf and stackb. The validator still lets digits through.

Level. Universal, _u-overview/feat-layers.md. A guideline; the validator was not changed.

Before (2.18). 2.18:docs/_u-overview/feat-layers.md:80: «We recommend that the layer identifiers consist of lowercase English letters».

Now (snapshot 24.09.2026).

• docs/_u-overview/feat-layers.md:80: «The layer identifiers consist of lowercase English letters». The other change in the file is a removed trailing space on line 261;
• docs/_oge/feat/: new pages Case-dsauf.md, Number-dsauf.md, Case-stackb.md, Case-sauf.md, Number-sauf.md. The text still has the old names: Case-stackb.md:14 — Case[stack2], Number-dsauf.md:15 — Number[sauf2]. About the pages themselves — PRs #1258–#1266 (June 2026);
• validator: tools/udtools/src/udtools/utils.py:60, layer — \[[a-z0-9]+\], i.e. digits are allowed. tools/data/feats.json has no layers with digits.

Evidence. diff of feat-layers.md; the listing of _oge/feat/; utils.py. In parallel the site renamed pages: Number[io].md → Number-io.md (_ka, _oge). This is infrastructure, not annotation.

What it means for English annotation. Nothing: EWT and the en registry have no layered features. For mova — a constraint on future own layers: the identifier is only [a-z]+, no digits, even though the validator would accept digits.

Registry. Not applicable.

Sources. docs/_u-overview/feat-layers.md, 2.18:docs/_u-overview/feat-layers.md, docs/_oge/feat/*, tools/udtools/src/udtools/utils.py.

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/dialects/ud-next/seeds/feat-layer-ids/