The motivation was to have a means of inserting an alternate implementation of, say, FOMEntry that implemented custom behaviors without having to create a new FOMFactory subclass. However, the approach is actually bit bogus (IMHO).
- James Garrett Rooney wrote: > On 7/14/06, James M Snell <[EMAIL PROTECTED]> wrote: >> Rob Yates did some performance testing earlier today on the latest code >> to find that resource use has jumped considerably. I suspected that the >> culprit was the alternatives mechanism I added a couple of weeks ago. >> After doing some investigation, that does appear to be the case. Given >> that Rob was the one who had the use case that needed alternatives, >> hopefully he'll weigh in on whether or not they're actually still needed >> :-). I'd love to back that change back out. > > What was the use case that prompted their addition? > > -garrett >
