Can I define some new languages, to be used in future code golfs? For example, CG-VB.NET would be a language which is equivalent to itself as VB.NET code, but wrapped in:
Module X
Sub Main
'CG-VB.NET code is put here
End Sub
End Module
And the same for several other languages such as C# and Java? What would I have to do to make it a real language? Is the definition here sufficient, or do I need to create a compiler? Formal definition?
I created Pyth a year and a month ago as a sort of pre-processed Python that would be competitive against golfing languages like CJam and Golfscript.
While it has since diverged away from Python, it still compiles into Python. Moreover, Pyth is now a widely accepted and popular golfing language with hundreds of answers on the site - and most of them aren't mine!
I think this is a fine path for golfers with a favorite language who want to compete for the shortest overall answer. Be warned that it takes a huge amount of time and effort, however.
In my humble opinion, such "new languages" are pointless; it's just another way to propose language handicap. The reason I say it's pointless is that it's not just the boilerplate that dooms languages like C# and Java for golfing purposes; they are just naturally verbose languages.
Given any Java or C# solution (even after stripping off boilerplate), I'm almost certain to be able to write a shorter version given the choice of GolfScript, Perl, or Ruby. (They are like my unholy trinity as far as golfing goes. A number of other contestants are also able to use J to great effect.)
We haven't really discussed the role and application of pre-processors (which is how I would characterize your suggested “language”).
But lets look at some near misses:
You should make rules specific to your puzzle to deal with language handicaps, for example, in Fibonacci function or sequence, they allow you to make a function or a sequence. In some puzzles, they only expect a function definition or even a code snippet, not a full class body with main method.
