Re: j2sdk build-depends cannot be satisfied?

[Stephen Zander <gibreel@debian.org>]

>>>>> "Stefan" == Stefan Gybas <gybas@trustsec.de> writes:
    Stefan> I have used Sun's definition of "runtime environment"
    Stefan> (i.e. core classes and JVM) in this case. And obviously
    Stefan> the maintainers of classpath and libgcj* used the same
    Stefan> definitions, otherwise their packages would provide at
    Stefan> least java1-runtime.

So classpath & libgjc *together* provide java1-runtime?

    Stefan> Of course, lib*-java packages should depend on the Java
    Stefan> core classes but IMHO not on something that contains a
    Stefan> JVM. Maybe it's better to call the virtual packages that
    Stefan> provides the core classes "java1-core" instead of
    Stefan> "java1-runtime" to avoid confusion?

What I think you're suggesting is that lib*-java classes would depend
on a virtual package that provides the java.* classes, e.g. classpath,
that was seperate from the virtual package that represented a complete
runtime environment, i.e java.* classes, JVM & tools.  That would mean
that kaffe, sablevm, and the various Sun jre, would then provide both
virtual packages.  Essentially:

        java1-runtime == java-virtual-machine + java1-core
        java2-runtime == java-virtual-machine + java2-core

Other than preventing the need for a dummy package to link classpath &
libgcj together (assuming my understanding above is correct), what
does this additional package gain for us?

-- 
Stephen

"And what do we burn apart from witches?"... "More witches!"