On Thu, Oct 11, 2001 at 03:07:25PM -0700, David Hull wrote: > Since yesterday I discoved that my patch to DBD-Oracle-1.12 to support > BFILE LOBs treats a non-existent BFILE as an error. This patch causes > them to be treated as if they exist but are of zero length. I tried > having the fetch function return undef instead, but this caused some > higher-level DBD or DBI code to report an error on that row.
Undef or error are the way to go. Surely non-existant BFILE's are significant and some people will want to detect that situation? Tim.
