[jsr294-modularity-eg] Pain

[Alex Buckley]

Peter Kriens wrote:
>> Regarding container-info.java and container-info.class as described 
>> below, they are outside the scope of a JSR. The 294 spec cannot 
>> mandate the existence of tools and the operational steps of a build 
>> process.
> Though I do agree JSR 294 does not mandate tools, it does define 
> concepts that are reified by tools. I.e. the javac compiler is implied 
> by the compilation unit.
> 
> Therefore, I fail to see why we could not (and I actually strongly 
> recommend it) define a container unit in the JLS that is reified by some 
> other tool than javac?

By your own description of how a container unit is used, it's not part 
of the Java programming language. It's purely an implementation detail 
for tools. The JLS is not a compiler or tool spec. In particular, the 
JLS does not imply the existence of javac. A command-line REPL that 
understood the CompilationUnit grammar would be perfectly valid if it 
implemented the static checks given in the JLS. (CompilationUnit could 
just as well be called SourceUnit.) In the Scala world, the REPL is just 
as important as scalac.

Alex
_______________________________________________
jsr294-modularity-eg mailing list
jsr294-modularity-eg at cs.oswego.edu
http://cs.oswego.edu/mailman/listinfo/jsr294-modularity-eg