That all sounds reasonable to me. Gary
On Mon, May 19, 2025, 20:04 Phil Steitz <phil.ste...@gmail.com> wrote: > The javadoc suggests that for both GKOP and GOP, this method is a no-op if > there is no capacity to add to the pool when it is invoked. But at least > for GOP, since the method calls create and create can wait up to the > configured borrow wait time for capacity to be available. When fixing > POOL-420, I changed the GKOP implementation to return immediately if there > is no capacity to add. > > I would like to > > 1. Change GOP so it similarly checks capacity and returns immediately if > there is none. > 2. Add addObject methods that take a wait time limit parameter. I don't > see much use for this in GOP, but I can imagine scenarios where users may > want to wait for keyed pools to subside when trying to add to another pool. > > Any objections? I am also willing to revert the change to GOP addObject if > others think this may break applications that depend on prior behavior. > > Phil >
