Hi,
now, that I have made myself more familiar with BeanUtils2, there are a
few things worth discussing regarding AccessibleObjectsRegistry.
Complex methods: I know, that BeanUtils2 is just an experiment, and that
the code was written to get it working fast for this purpose. But some
methods are really hard to read. Take for example resolve() on
MethodsRegistry. There are at least two things happening: 1. try to find
a direct match, 2. if no direct match can be found, start a some how
more complex search. I think this could be separated into several
methods. If you agree, I can create an issue and write a patch.
Magic numbers: I really don't understand, how object transformation
costs get calculated in MethodsRegistry.getObjectTransformationCosts.
What does 0.25f mean? And why do we add 1.5f? Are this values you came
up with from the development of BeanUtils1? If so, this should be
documented in the code.
Terminology: I really don't like the name of the class. My opinion is,
that the term "object" should never be overloaded. The problem is, that
AccessibleObjects refer to methods and constructors that are accessible
from the caller (e.g. public for a caller outside the containing package
of the bean class). Now, if you are new to BeanUtils2, your first
thought might be, that those registries are holding objects, that are
accessible themselves.
What I'm trying to say is, that maybe the name should be changed to
something more convenient, for example "AccessibleOperationsRegistry".
Since the general definition of a class is, that it defines the data and
possible operations on that data for a set of similar objects, I think
that name would be very appropriate (AccessibleObjectDescriptor, etc
should be renamed likewise).
all the best
Benedikt
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
For additional commands, e-mail: dev-h...@commons.apache.org
