textlab/v0 package

Package textlab runs every string algorithm in p/moul over one input at once, so they can be compared rather than read about one at a time.

Each of these has its own demo realm showing it alone. What none of those can show is how the answers differ for the SAME word, which is the only question a reader actually has: soundex and levenshtein both claim to tell you whether two words are alike, and they are measuring different things.

soundex keys on the FIRST LETTER and then on consonant classes, so it is blind to how the rest is spelled. levenshtein counts edits and is blind to how it sounds. "knight" and "night" are the clean case: one edit apart (83% similar) and yet soundex calls them unrelated, K523 against N230, because the silent K changes the first character of the code. "robert" and "rupert" go the other way, identical codes at two edits apart. Measured 2026-09-23 against the packages in this tree, not quoted from a textbook.

1/r/nym-moultest123/x/allinone/textlab/v0:knight
2/r/nym-moultest123/x/allinone/textlab/v0:knight/night   compares two

Eight packages, no state, nothing to redeploy carefully: the input comes out of the render path and nothing is stored.