Alexandre Benson Smith wrote:

>I know one of the best software engineering practice is to reduce code
>complexity, and I take this as the first rule to follow, since simple
>code means less bugs, less time to understand and perhaps some other
>bennefits. So I do just what is needed nothing more.
>
>What I'd like to ask you is:
>
>What about if FB has two character representations ? 32 bits Unicode and
>16 bits Unicode. At database creation the type of charset to be used in
>storage was choosen between the two kinds.
>
>Why I ask it:
>1.) Who needs only 16 bits unicode will not waste space (disk reads)
>2.) Who needs 32 bits will have an option.
>
>Don't know how dificult is to teach the engine internals to handle 2
>diferent charsets (from what I understand from your messages, this will
>be just "some" funcions).
>
>This could be a case where implementing more complex code could lead to
>a better sollution ?
>
>
>
>

You have to recoginize that there is almost no different between two
internal character sets and an infinite number. If every copy or
compare operation has to check for conversion, we end up with what we
have now.

So I'm going to ask the question again: What benefit, if any, does 16
bit Unicode have over UTF-8 for internal representation?