The code is in pretty decent shape. The one unit test I have validates
at least a basic level of funcionality. It's not too surprising given
that this was mostly a migration of code from glassfish to grizzly.
I've refactored the code where I feel confident I'm not going to break
things and it's cleaned up rather well in places. There are a few
shortcomings, though, that we'll need to address over time. Most of the
logic is really centered out http handling and mapping servlet contexts,
etc. This again, isn't surprising given then history of the code. It
also makes assumptions about the type of the Adapter used and there
currently is no way to define the adapter in the configuration file.
This makes expanding the functionality a tad limited but shouldn't be
too hard to adapt.
Overall the code is bit more convoluted in places than I prefer but it's
been pretty focused on one task for a while and we'll have to work on
generalizing things and abstracting out what we can. I had hoped to
duplicate some of the "pure" grizzly tests using the config approach but
things like that adapter issue make it a bit less obvious as to how to
proceed on that front. That said, we can still build plenty of test
cases otherwise. A nice small web app deployed under this would be
interesting. For now, I'm going to focus on glassfish integration. I'm
going to create a branch in the glassfish tree and begin retooling
things to work with the new schema. That's probably as an extensive
stress test as we'll get in the short term so that should help shake out
more oddities.