[Jsr308-statements] First draft of proposal

[Eugene Kuleshov]

Hi Trevor,

  It is really great summary.

  I have few notes in regards to the bytecode representation.

-- First of all there is no need to have separate "visible" and 
"invisible" attributes. For regular annotations those are used to 
differentiate between attributes available trout reflection API and 
those that aren't. Since we aren't planning to expose statement-level 
annotations as well as other code-related annotations trough reflection 
(btw, it worth to state that in the proposal document) we don't need 
"visible" attribute.

-- It seems like we can just use another target_type and special 
reference_info within the same RuntimeInvisibleTypeAnnotations 
introduced in base JSR 308 proposal. Though that attribute could 
probably named better for such purpose.

-- Use case from "Object Allocation" section (on object creation, new) 
is covered by the main JSR 308 proposal

  I've also added some couple more examples to the wiki (fine-grained 
statement advices for AOP and example from [a]C#)

  regards,
  Eugene


Trevor Harmon wrote:
> I've completed the first draft of a proposal that would supplement 
> JSR-308 by describing how statement annotations can be supported. It's 
> available here:
>
> http://doc.ece.uci.edu/mediawiki/index.php/JSR-308_Statements
>
> Note that it's a wiki, so feel free to add to it and revise it. (As a 
> protection against spam, you'll need to create an account before you 
> can edit.)
>
> The main thing it offers is a collection of all the use cases for 
> statement/block annotations that have been mentioned in the mailing 
> lists. I suggest that any new use cases we come up with should be 
> posted to this wiki instead of (or at least in addition to) the list. 
> That way, we won't "lose" any good use case ideas among the flurries 
> of email that we get.
>
> Trevor