Bryan Pendleton wrote:
Given there is a well documented wire protocol (DRDA) why would we
invent a new different wire format for some commands?
Following an open standard is a good thing and does allow other
parties to implement clients against Derby.
Well put.
I think I worded my message too strongly. Let me try again:
Given that:
- our client talks only to our server
- there aren't any freely available compliance tests for DRDA that
I know of
- most of the community doesn't know much more about DRDA than what
we read in the spec
How would we know whether we had implemented the wire format correctly
or not?
By people reviewing the code, by the coders including references to the
spec when coding, by finding out by someone looking at the code later
(maybe the incorrect implementation clashes with a new correctly
implemented command), by someone executing a non-Derby DRDA tool against
Derby, etc.
That is, if we give our best shot at implementing this new SQLSTT
verb, but we don't get the wire format exactly right, what would be
the actual implications for us?
I think it would be the same as if we implemented any other spec
incorrectly. E.g. LIKE with language based collation was changed to
match the spec.
Dan.
