Issues GitHub issue

vibecode
{"vibecode": {
    "doc": "uma-history-issues",
    "role": "historical issue list from the prior Ruby Uma project: open items about canon.json tag coverage, root/head structure, attribute semantics, and content-model constraints",
    "key_concepts": ["ruby_uma_issues", "canon_json_gaps", "html5_content_model",
        "tag_coverage_gap"],
    "status": "historical"
}}

Open issues GitHub issue

Expand Canonical Tag Coverage To Match Runtime Spec GitHub issue

lib/uma/canon.json currently defines only a subset of tags in tags. A direct comparison against lib/uma/tags.json shows 70 tags present in the runtime spec but not represented as canonical tag definitions (examples: article, section, nav, figure, figcaption, fieldset, label, audio, video, source, track, template, style, noscript, dl, dt, dd, ruby, rp, slot).

Concern: the normalized source of truth cannot fully regenerate the runtime tag surface without additional canonical tag entries or canonical rule sources for these tags.

Correct Canonical Document-Structure Rules For Root Tags GitHub issue

lib/uma/canon.json models html.children as body, link, meta, title, and script, but does not include head. In HTML5, metadata elements (link, meta, title, script in this context) belong in head, not directly under html.

Concern: denormalization from canon can produce structurally invalid parent/child rules unless the root/head model is corrected and expanded.

Add Canonical Metadata For Missing Attribute Semantics GitHub issue

The canonical file includes reusable attribute groups (default, remote, form-fields) but omits many tag-specific attributes that exist in lib/uma/tags.json (examples: input.accept/min/max/step, textarea.wrap, select.multiple/disabled, table.border/summary, media attributes like video.poster, audio.controls, and head metadata attributes like meta.name/content/http-equiv).

Concern: canon does not yet capture enough normalized attribute semantics to reliably regenerate the richer runtime schema for many HTML5 elements.

Encode More HTML5 Content-Model Constraints In Canon GitHub issue

Current canonical sources (blocks, inlines, most-inlines, list-items, etc.) are useful but broad. They do not yet encode several important HTML5 constraints, such as specialized groups and restrictions for: - table internals (caption, colgroup, sectioning and row/cell ordering nuances) - form composition (fieldset/legend, label constraints, optgroup in select) - interactive-content nesting restrictions and other invalid descendant combinations

Concern: without these constraints in canon, denormalization may over-permit invalid structures or under-specify valid ones.

© 2026 Puck.uno