These threads on the struts-user mailing list just go to show how tricky a problem persistence frameworks are. Not strictly technically speaking, but more of a personal sense. I see persistence in a class of problems that are: a) extremely common, b) fairly simple to understand, c) fairly simple to code a framework that functions. (See also: anything to do with databases, templating, web application frameworks, code generation.)
The problem (if you can call it that) is that everyone creates a persistence framework. They're not really different from one another except in the way your favorite pair of jeans feels better than any other -- a distinction without a difference as Barb likes to say. But you get used to your way of doing things, even if it's not more functional than another way, your jeans start getting frayed at the cuffs. So you stick with your homegrown solution that has some bumps here and there, and one hierarchy of code has some nasty funk emanating and needs to be refactored, but there will be time after this project is wrapped up...
This is a kind of hubris run amok. I don't know if you can point to a particular cause, although if you were an armchair psychologist you'd probably generalize about solitary programmers being big fish in small ponds. But you'd be wrong just as often as you're right.
There is a certain pleasure in finding a tool that fits nearly all your requirements and modifying it meet the rest of them. It gives you a focus that you may not get with your home brew, since you know you can't change everything. Broken windows are still an issue, but instead of feeling the burden of direct responsibility -- another item for my TODO list! -- you can simply point it out to the project leader, who knows the project better anyway and has a better eye for prioritization than you do. After all, don't you have plenty of mountains to climb as it is?