☰
2.1.1: the adoption checklist - deleting the hand-rolled panel is what breaks pages, not mounting the module
2.1.0: onToggle(open, source) so graduation cannot feed back into itself; toggle tooltip swaps with state; document pas…
2.0.5: module version generated from the pack version; selfCheck(panel) argument documented; node --check and sha256 ad…
2.0.4: selfCheck accepts declared OR snapped border width (failed correct panels at dpr 1.5); serve-as-a-file and same-…
2.0.3: record the packaging trap - a doc will publish happily without its payload
2.0.2: add the module (and the three action icons) to the files allowlist - 2.0.0/2.0.1 shipped the doc without the pay…
2.0.1: description now leads with the module and matches the spec (the old 'edge-to-edge band' wording is what pup buil…
2.0.0: ship the newbie panel as ONE shared module every dashboard mounts, instead of a spec seven apps each re-implement
1.41.0: phrases must fit the cell at 744 (trigger noun early); measure the span not the cell; dead-vs-losing selector
1.40.1: publish your arithmetic, not just your verdict
1.40.0: 700 settled by experiment (hold width fixed, vary only layout); layout dominates width
1.39.0: a structural audit cannot see a missing function; assert behaviour and self-check for an empty panel
1.38.2: the reasoning still argued for 620 while the rule said 700; they now agree
1.38.1: the content 1.38.0 claimed (page padding is contract; specified-value read means unrendered pane)
1.38.0: page padding is part of the contract (identical CSS gives different columns otherwise); a specified-value compu…
1.37.0: a shared .btn class can eat the demo fill by source order; assert structure because the browser repairs malform…
1.36.1: retract the three-pane-widths table; it was measured on unforegrounded tabs, and every dashboard reports 744
1.36.0: there is no single pane width (632/744/851 measured simultaneously); 700 serves the range
1.35.0: delivered_to is an array; mocks built from your own assumption cannot falsify it; test sends use a no-op prompt
1.34.0: breakpoint 620 -> 700; pane width is not viewport width, and one readable column beats two truncated ones
1.33.0: a timed-out send may have delivered; never re-send, and never drive a send through a transport with a shorter t…
1.32.1: grep max-width is not a breakpoint list; grep @media and walk the widths
1.32.0: minmax(180,220) can never reach 180 (min-width:0 on the fr track); use clamp(180px,30%,220px)
1.31.0: viewport report names offenders; a harness matrix is measurement, not verification
1.30.0: un-park by DELETING the branch (a flag hides an unexercised path); assert you are on the user's box in code bef…