>> I've mentioned it before, but I really think it would nice to decouple the
>> release cycles of the core engine from the interfaces. Make them separate
>> projects.
> We have sort of done that now by pushing everyone to the CVS version.
This makes some sense to me for ODBC and JDBC, which are large and
complex enough to deserve the label of separate projects; and moreover
they go out of their way to work with multiple server releases. I'm not
sure it's appropriate for any of the other interface libraries, though.
IIRC, at one time ODBC *was* a separate project, and we decided that
that wasn't working too well. Anyone recall the reasons we pulled it
into the main CVS tree? Wouldn't do to make the same mistakes twice...
regards, tom lane