a separate file, testsuite.js, loaded dynamically by Sam.

Sam::cmdTestParse(),

• loads code dynamically read db, test parse, compare the results, format the output
• pulled from Noam

dict g=o, syllables

• one test word for each of 28 rules. rules fall into three categories: endings, tones, consonant clusters
• one test word for each leading consonant
• one test word for each vowel pattern
• one test word for each final consonant
• one test word for each tonemark
• one test word for each class/tonemark/vowel-length combination
• all words with rule “excp”

dict g=m, multi-syllable words

dict tlm=m,

• all words with manual translit

dict cpm=m.

• all words with manual parse

• strings with punctuation
• strings with speakers
• strings with spaces

• create a special test story that tests all strings and all words
• no good, too difficult to isolate the errors for diagnosis