The canonical comparison text: the entviz's top label in square brackets,
then the cells' text in grid reading order (left→right, top→bottom),
space-separated, with each blank cell preserved as a ·. Case-exact — this
is the read-aloud verification surface, never localized.
The label is REQUIRED, not decoration. The cells carry the core alone, so
anything the parser folds into the fingerprint instead of the cells — a
did:<method>:, a urn:<nid>:, a SWHID/gitoid object-type, a bech32 HRP —
is invisible here, and two values differing only in a folded prefix produce
byte-identical cell text. Affirming identical from cell text alone was
therefore affirming sameness between values the engine itself calls
different: did:good:X versus did:evil:X read out the same. The spec
requires any read-aloud or comparison-text procedure to include the label
strip's prefix (spec.md, the text channel's folded-prefix exception); the
whole label is used because it is what the picture actually shows, and it
additionally carries the encoding, the size, and any presentation prefix
(0x, a multihash header) that the cells also drop.
The label is rendered WITHOUT a line budget, so it is never elastically
truncated. A comparison text must be a function of the value alone; folding
in the grid width would make the same value produce different text at
different font sizes or aspect ratios.
The canonical comparison text: the entviz's top label in square brackets, then the cells' text in grid reading order (left→right, top→bottom), space-separated, with each blank cell preserved as a
·. Case-exact — this is the read-aloud verification surface, never localized.The label is REQUIRED, not decoration. The cells carry the core alone, so anything the parser folds into the fingerprint instead of the cells — a
did:<method>:, aurn:<nid>:, a SWHID/gitoid object-type, a bech32 HRP — is invisible here, and two values differing only in a folded prefix produce byte-identical cell text. Affirmingidenticalfrom cell text alone was therefore affirming sameness between values the engine itself calls different:did:good:Xversusdid:evil:Xread out the same. The spec requires any read-aloud or comparison-text procedure to include the label strip's prefix (spec.md, the text channel's folded-prefix exception); the whole label is used because it is what the picture actually shows, and it additionally carries the encoding, the size, and any presentation prefix (0x, a multihash header) that the cells also drop.The label is rendered WITHOUT a line budget, so it is never elastically truncated. A comparison text must be a function of the value alone; folding in the grid width would make the same value produce different text at different font sizes or aspect ratios.