full-text searching using rdbms

I posed this question on the sybase board, but the responses were not in favor of using sybase. They were a bit more favorable towards Oracle, however, so here I pose the same question again.

I've read about full-text searchers being implemented using some sort of rdbms. I was wondering if anyone has coded such a beast using sybase or any other rdb and would like to share with me any experiences and performance results of the system.

More specifically, I'd like to know whether you used an index and pointer list? I may have heard that support for these structures is already built into whatever Oracle provides. Is this true?

If not, then was the index list stored in a single table? A row for each word, then the pointer list?

I guess whatever means was used was sufficient, in terms of performance and disk overhead, for whatever amount of data you were storing, then? What sorts of things did you do to optimize searches? Indexes? Indexes for things like all words starting with the letter a? b? c? even ab? How far did you go? and possibly how much was built into Oracle?

Does sql or Oracle support searches of the form:

select title
from indexlist
where keyword = s*

where it'll match the keyword field data w/ all words starting with s for instance?

Were documents actually stored in the database or maybe just a pointer to the actual file located on disk somewhere? Why?

If you haven't already guessed, I'm new to databasing and full-text searching, so ptrs to good references will also be appreciated.

Thanks for any info you can provide me with.

Jae