Skip to content

data(WO87): evidence-based rename map + defect register - #360

Merged
ronaldtse merged 1 commit into
mainfrom
docs/wo87-evidence-and-map
Oct 24, 2026
Merged

ronaldtse merged 1 commit into
mainfrom
docs/wo87-evidence-and-map

Conversation

@ronaldtse

Copy link
Copy Markdown
Contributor

The Phase-2 evidence and its consequences:

  • Evidence: all 33 ids decoded through the reference loader (28 documented; 5 load errors are findings themselves).
  • Renames (evidence-backed): urd-g2p→urd-ipa (narrow IPA + stress), fas-g2p→fas-phonemic (Latin-letter phonemic), tha-g2p→tha-phonemic (toned phonemic family). Old ids alias at the next index tag — zero breakage. The diac/latn names verified target-accurate — unchanged.
  • Defects found by the sweep: ara-diac-news-1.1 artifact LOST (release never existed, bytes gone — withdrawn, README corrected); 4 plane-family zips violate the IMF v1 contract; ara-diac-small-1.0-int8 emits spurious tokens (superseded line).

Target documentation for 28/33 ids (5 load errors are findings, not
gaps). Renames (evidence-backed): urd-g2p->urd-ipa (narrow IPA with
stress), fas-g2p->fas-phonemic (Latin-letter phonemic),
tha-g2p->tha-phonemic (toned phonemic, joins the family). diac/latn
names verified target-accurate — unchanged.

Defects: ara-diac-news-1.1 artifact LOST (withdrawn; README corrected),
plane-family zips violate the IMF contract (x4), small-1.0-int8 emits
spurious tokens.
@ronaldtse
ronaldtse merged commit ffd26cf into main Oct 24, 2026
12 checks passed
@ronaldtse
ronaldtse deleted the docs/wo87-evidence-and-map branch October 24, 2026 06:40
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant