JEP-118: Why are parameter names not mandatory?

[Alex Buckley]

On 5/1/2013 3:18 PM, Jesper Steen Møller wrote:
> I'm currently examining these two JEPs, and was curious: Why not make
> the ParameterNames attribute mandatory at least for non-private
> methods. I don't quite get why it makes sense to add this info which
> will be useful for introspection, IDE's and annotation processing,
> but only make it optional - even defaulting to 'off'?

Please see the introduction to chapter 2 of 
http://cr.openjdk.java.net/~abuckley/8misc.pdf.

The bottom line is that for every programmer you please by including 
parameter names by default, you intensely anger another. Some people 
dislike the extra class file size; some dislike the new compatibility 
surface; some regard it as exposing sensitive information.

So you need an opt-in mechanism. If you want it in the Java language, 
you need to define syntax and scope. See the first thread at 
http://mail.openjdk.java.net/pipermail/enhanced-metadata-spec-discuss/2012-November/thread.html. 
Personally I would focus on simplicity (@ReifyParameters on a type decl 
means store p-names of public methods) but that did not meet others' 
needs. For once, I felt not standardizing was the right answer. This 
freed us to focus on getting the interaction right between 
java.lang.reflect.Parameter, type annotations, and repeating 
annotations. I hope to standardize the opt-in mechanism in SE 9.

Alex