split to words and db in Ram

[John W. Miescher]

Thanks for the explanation. I find splittowords indeed very useful in
conjunction with my proximiy search alias soundex routine (which is
absent from Valentina) because it pre-sorts and also removes duplicates
faster than RB would. In a subsequent step I build a db and fldstring as
per your example except I do it in RAM:

  db = new VDatabase(EVStorageType.kRAM)
  db.CollationAttribute( EVColAttribute.kStrength ) =
EVColAttributeValue.kPrimary

but the CollationAttribute does not seem to have any effect here, i.e.
accents are NOT ignored. Why?

rgs, john m.