Fwd: RFC: developing DBD::SQLite

Amalgamation what the SQLite maintainers recommend, and it's the
most portable, etc.  So this is the right choice.

Also, given the way SQLite is maintained (backward compatibility is fairly
strict, they jump through hoops in their API to make it so), bundling
is fine, and is intended by the designers.

I agree it can be a slight hassle to update to the current
amalgamation, run tests, and then release on CPAN, but that
cycle can be made pretty short.

Having the *ability* to use installed libraries... say on a system
where the installed libs are newer/threadsafe/being tested, whatever,
is a useful feature.  But, in my experience, it was only
needed because the DBD::Sqlite *wasn't* being well maintained.
If it had been, I would have just used the new DBD::Sqlite and
never worried about it.

The DBD::SQlite should use

  a) already installed libs if *available and compatible*
  b) *bundled* amalgamation, which is guaranteed to be available and compatible

That's it.