Knut Anders Hatlen wrote:
"Kristian Waagan (JIRA)" <[EMAIL PROTECTED]> writes:
Committed 'blobframework_v2.diff' to trunk with revision 529185.
Very nicely commented patch Øystein!
OK, so I'm more than three weeks late to review the patch... :)
Anyway, I came across CallableLocatorProcedures.java and a couple of
questions came to mind:
A) It seems to contain a lot of almost identical code for CLOBs and
BLOBs. Would it make sense to reuse some of it instead of duplicating
it? For instance, blobCreateLocator() and clobCreateLocator() have
exactly the same structure, and could have been written as a generic
lobCreateLocator() which took a CallableStatement, the SQL text and the
holdability as parameters. blobCreateLocator() and clobCreateLocator()
could be just wrappers are around the generic method. The same comment
also applies to [bc]lobReleaseLocator, [bc]lobGetPositionFromLocator,
[bc]lobGetLength and [bc]lobTruncate. I think this could reduce the size
of the file considerably, and it would also be easier to maintain it.
I agree that this should be considered. When I created the file there
was not much logic except calling the underlying stored procedures, and
the most straight-forward way of doing this was to start with a list of
methods to implement and just use a macro to generate code for all the
methods.
However, later more logic has been added and I agree that it would be
good if that logic could be shared. The ironi is that the those "sister
methods" that have most of "extra logic", cannot share implementation
since they operate on different data types (byte arrays vs. String).
I will put rewriting this class on my list, but I need to prioritize
getting locators to run without failures first.
B) Why does blobCreateLocator() get the holdability from the connection,
whereas clobCreateLocator() always uses non-holdable cursors?
This kind of proves your point. :-) I do not see any reason for this
difference. Holdability would not matter here, anyway.
--
Øystein
