Re: Database I/O Modernization

On Mon, Sep 11, 2017 at 10:34 AM, Charles Wilt <charles.wilt@xxxxxxxxx>
wrote:

​And that's the problem. It adds a layer without adding any value. In
fact, it decreases value (by increasing complexity & interference) vs.
just using RPG op-codes in the layer above.

Good point about decreasing value, Charles. For example, the interface
provides only one "open mode" for a file. Rather than alternatively
allowing open for input, or open for output, open for overwrite, or open
for append, or open for update, depending on the requirements of the
application, the procedure interface requires an open mode that
accommodates every possible combination. That in turn requires a lot of
additional code to enable all the combinations of record locking that may
be required for various types of reads (chain w/ lock, chain w/o lock, read
previous w/lock read previous w/o lock, read next w/lock, read next w/o
lock).

Some applications may require read first record, read last record, various
reads with keys equal, or keys greater than, or keys less than, or reads
that involve a partial key rather than the complete key... consider all the
lock / no lock options and all the extra "actions" (I'm referring to the
action parameter in the procedure interface) that would need to be defined
in order to accommodate that level of flexibility.