John Cowan wrote:
> That's the whole idea of pluggable database libraries. All you have to
> do, minimum, is to write code to answer the questions "is the string X a
> member of the datatype" and "do the strings X and Y represent the same
> member of the datatype". You can also support your own parameters to
> provide restricted subtypes.
OK, I've done it now, but is that really the minimum? So far it seems I
also need to:
1. Compute hash codes for the strings
2. Buffer the strings from streams, or write code that validates the
streams directly
3. Write a factory that finds the right library
4. Write the library that creates the right data types
5. Install the factory (Still haven't figured out how to do that one)
6. Provide an object representation of the type
It's doable, but it's a little more complex than I expected, even for
simple cases. I thought I could get away with just writing a method that
took a string and returned true or false, but there appears to be a lot
more to it.
--
Elliotte Rusty Harold
XML in a Nutshell 3rd Edition Just Published!
http://www.cafeconleche.org/books/xian3/
http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim
