Brian McCallister wrote:
For Date and its ilk, and arrays, providing special handling for them is reasonable. Arrays of immutables will be a pain, but then so is caching of LOB's (should we even cache LOB's?)
If, then we should only cache the LOB itself. Ie the locator object and never the data it points to.
This is current behavior of copy in FieldTypeClasses$BlobType (+Clob, UDT's).
Following the LOB and caching all data is like dereferencing a void* in C (you don't want to go into the void). ;)
Agree caching LOB objects doesn't make sense. Think the user is resonsible to avoid this by excluding these objects from being cached.
This is the problem in current implementation of Blob, Clob handling (JdbcTypesHelper$T_Blob), we always return Blob/Clob value instead the Blob/Clob locator object.
Armin
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
