Files
adr-sml/docs
Pouya LajevardiandClaude Opus 5 ca1c2524e1 fix: footer email reflow to zero under all four methods; reopen the skip-link residual as Q61
The footer mailto was the last recorded overflow: `info@smlcompany.ca` demanded
310px of min-content in a 224-243px column. One declaration —
`overflow-wrap: anywhere` on `.footer-contact a[href^='mailto:']`. `anywhere`
and not `break-word`, established with a negative control rather than from the
rule: only `anywhere` reduces min-content, and `break-word` injected in its
place failed the same 88 rows as the unfixed baseline.

Minimum-font-size 32 goes 88/352 -> 0/352. All four methods now read 0 of 352
(22 pages x 16 widths). Normal-settings identity: 0 differences across 8
metrics, with a positive control proving the comparison can detect one. It also
closed 57 element-level cases the page-level table reports as clean, hidden
inside `.wrap`'s 96px gutter.

The item-2 ruling is NOT applied, and this is the deviation to read first. The
acceptance rested on "no CSS mechanism can see minimum-font-size", which is
false: the font-metric units — `ch`, `ex`, `cap`, `lh`, `rlh` — read the used
font size and double, in property values, in `@media` and in `@container`. Only
`rem`, `em`, `ic` and `px` are blind. And the cost is not a convenience loss:
keyboard focus lands entirely behind the opaque header on 290 of 1,455 stops,
36 of them inside `#main`, which is WCAG 2.2 SC 2.4.11 at AA — the same level as
the 1.4.10 failure it was traded against. A build of fce89d4~1 measures 0, so
the header fix created it. Opened as Q61 with a verified candidate; docs/06
restored to unticked.

R20's gate is now a build failure rather than three prose cross-references,
which demonstrably did not gate it: with two articles published the build and
all five checks passed while both header defects shipped. SiteHeader throws.

R11's two majors (@astrojs/mdx 7->8, typescript 6->7) move to a new cutover-prep
group in docs/06 with the 19-pin currency sweep.

Two review rounds, eight findings, all resolved; four of round 2's five were
defects in round 1's own fixes. Stopped at two per D19.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Md3GndFqWPzK78xAoebsg5
2026-09-01 14:51:37 -04:00
..