Virtual extension methods -- a strawman design

[abies at adres.pl]

"Brian Goetz" <brian.goetz at oracle.com> napisał(a): 

 > As you correctly point out, the choices are VM-only and VM+compiler, and 
 > compiler-only is not an option.  I see pros and cons for both options.  The 
 > VM+compiler option is obviously more work for us, but seems likely to reduce 
 > the impact on the tools ecosystem.  Ignoring the "which is more work" aspect, 
 > can you outline what you see as the advantages of the VM-only approach?
 > 

Consistency - you will always get current target of default invoked, avoiding const-like inlining.

As far as tools are concerned, some modification is anyway required to understand the extended interface format. Question is, does supporting two distinct versions of method 'guarding' is really helping to reduce the impact on tools? Will the compiler approach be enough for some tools to be compatible with really minimal changes? I'm bit scared that it will help in small way in short run, but require considerable more effort for tools like smart IDEs in longer run (where they will have to 'decode' compiler tricks to understand that in fact guard method is used).

Regards,
Artur Biesiadowski