[jsr294-modularity-eg] Pain

[Alex Buckley]

Richard S. Hall wrote:
> The "simple as possible" part for visibility is a little more 
> complicated. This is where we get into the "requires" concept. Since we 
> couldn't agree on all the features needed for "requires", we decided to 
> leave it completely up to the module system being used. This doesn't 
> seem "as simple as possible".
> 
> This brought us to Peter's proposal which, ignoring all the details, in 
> my view ultimately boils down to defining a simple "requires" definition 
> that is agreeable to and can be mapped onto various module systems. 
> Thus, this one simple "requires" concept could be taught to all users 
> and they could get the same behavior no matter which module system they 
> used.
> 
> If possible, this now seems like a step in the simpler direction, even 
> if we only define this simpler "requires" concept to just provide 
> visibility to all public types in the target module and nothing more.

I'd like to hear if Peter agrees with your summary, but I have no 
problem in agreeing to this definition of 'requires'. The 'target 
module' would of course be chosen by a module system up to versioning 
and other details. This was, substantially, requirement #6 back in [1].

Alex

[1]http://altair.cs.oswego.edu/pipermail/jsr294-modularity-eg/2009-January/000187.html