RE: New Diagnostic Code Requested

> -----Original Message-----
> From: Robert Waldstein [mailto:wald@library.ho.lucent.com]
>
> QUery by example:
>     - So does a with umlaut translate to a
>     - diphthong (ae)         translate to ae (a followed by e)
>     - oneHalf (1 over 2)    translate to  1/2
>     - a superscript 2       translate to  a 2
>     - capital A            translate to "a" (guess we handle 
> this with an attribute, do we do the others?)

Bob, I think most of your examples fall into what I would call
normalization, not translation.  The character AE-ae is not the same as an a
followed by an e.  If the target characterset does not contain the AE-ae,
then a diagnostic to that effect should be sent back and the client/user can
decide whether to turn the diphthong into an "ae".

Now, we're heading into tricky ground here.  I do all sorts of
normalization.  I shift case, I pull apostrophes out of words, I throw away
punctuation.  I throw away diacritics.  Some of that behavior is going to
have to change as I support more non-English data.

Ralph