[wtp-dev] Source Quality 2: serialVersionUID -- please don't make me fix

[David M Williams <david_williams@xxxxxxxxxx>]

Another common problem that causes warnings
in our source code is failure to declare a serialVersionUID for
a serializable class. 

The full warning looks like this: 
The serializable class DoubleObjectPrivilegeImpl
does not declare a static final serialVersionUID field of type long

This generates about 150 warnings in
about 15 projects. 

Technically, classes which are serializable
really should declare this static field since if that class ever was serialized,
then any
future (or coexisting) versions might
have to know the serialVersionUID to be able to deserialize it correctly.


Granted, this is the kind of warning
that *might* never be a problem, as many classes are serializable for trivial
reasons, 
and would never be appropriate to actually
serialize them .... but that is the 'contract' of serialization! (And,
often you'd never really 
know it was a problem until that future
version existed, or someone started doing something over a network, 
with different VM's on each end! ...
which is, I suspect why its often overlooked, and why they give a warning).


My main concern is that we have so many
of them, it makes it hard to see any "real" warnings. 

And the cure is so easy .... just use
QuickFix to add "default" versions. 

         
/**
         *
Default
<code>serialVersionUID</code>
         */
        private
static
final
long
serialVersionUID
= 1L;

(I think you'd only have to use the
"generate" version of the quick fix if you really had some concrete
reason to, such as you 
had existing serialized versions you
were trying to retrofit). 

I know many of these come from EMF generated
classes. Is this an issue? Something that EMF needs to address better?

Thanks (either for fixing, or explaining
why it can not or should not). 

Remember, our goal is zero warnings
during our builds.