Is there any mileage in looking at SQLJ for the programmer's view of the API? The burden could then be placed on the IDE to pre-compile/compile the SQLJ down to a particular implementation, whether RMS, an API over RMS, some future lite JDBC, an in-memory DB, a commercial database (Sybase Ultralite, Oracle Lite, Pointbase, ...) etc. If I read the specs correctly, SQLJ is supposed to be interoperable with JDBC in the same application (and even that could be ignored for ME needs, if we chose to go beyond the standard) which in practice means sharing connection handles and cursors, I think. The limited J2SE/EE SQLJ implementations to date have taken the easy option of pre-compiling to JDBC, but I don't know of any *requirement* to do it that way. regards, Carl Zetie |