I need your opinion...
rssh at gradsoft.com.ua
rssh at gradsoft.com.ua
Sun Mar 22 00:48:50 PDT 2009
>>
>
> I mean logic not name.
> And this is nothing similar with traits (lisp I do not know).
> Glue classes you can 'bind' to few classes at once.
>
May be. Different people think in different way.
For me, glue classes is 'traits' (i.e. compositivie units of behaviour)
mixed with 'extension methods'. I. e. reading you proposal I think 30
seconds, that say 'o, traits with extension method'. If word 'like traits
wich can be bound to classes as extension methods' was in the first
paragragh that time to understand such proposal for me, was 3 sec. instead
30. But this is only as my head work (I can tell you what is hard to
understand only from my introspection, sorry). Other people think in
different way, so may be functionality of glue classes is bound to such
terms only for me. Other people will tell you what hard for them.
Yet one thing, which can help: add more concrete example and describe
in detail process of method call (if method call is changed on JVM level)
or compiling to traditional java source code (if method call does not
changed on JVM level).
For example, I can't understand - how JVM or compiler will locate trait
with such function signature during method call ? By building dependency
'glue dependency graph' and collecting candidated in all possible glue
classes ?
More information about the coin-dev
mailing list