David W. Van Couvering wrote: > I'm not sure I understand this logic. Shouldn't we be providing useful > error messages with enough detail as possible? Standard error messages > may be reusable, but they may not be helpful. Are there space concerns?
Space concern is one factor. The other is that approaching this on an ad-hoc basis doesn't seem the best approach to me. Will lead to lots of different messages written by different people that really are describing the same situation. For example with your proposed fix would would have had a message just for CLOB that really isn't in line with the style of the other messages. Why not address it for BLOB as well, what about other data types, what about the corresponding client message? I've seen this before, the data type messages used to each print the type in their own way, much as the text of your message proposed. Thus we had no consistency, Clob vs. CLOB, DOUBLE vs. DOUBLE PRECISION vs double, DECIMAL vs NUMERIC vs. DECIMAL/NUMERIC. I'm sure we could write better error messages, but let's make a consistent effort around that, not little fixups in unrelated patches. If the standard message for value out of range is confusing, what could be done to improve it for all cases? Dan.
