Dave Cramer wrote:
> Laszlo Hornyak has finished off the proof of concept patch that I
> submitted a few months ago.
Any reason you wrote your own scan-the-whole-pool LRU rather than using
something like LinkedHashMap?
I didn't dig into the code too closely but it looks like you are using
the statement object directly with no wrapper. Doesn't this run the risk
that you will resurrect a previously-closed statement? Normal statement
objects have a one-way lifecycle, once they are closed they cannot be
resurrected, if app clients have a reference to the real statement then
potentially they'll see different behaviour when the statement starts
getting reused. That smells dangerous; not because any sane application
will rely on it, but because it will be a source of very hard to find
bugs. (e.g. it's fairly common and harmless to close an already-closed
statement.. but that's suddenly disastrous if the statement has actually
been pooled & reused in the meantime)
Are all the abstract/interface/concrete-implementation classes strictly
necessary to get it operating under multiple JDBC versions? They are a
real pain when it comes to maintaining the code. For example I can't see
offhand why you need to do that with StatementPool.
-O
