From rPath Wiki
- How do we split out upstream byte code from gcj byte code?
- Use flavors. Right now gcj means built from source with gcj and !gcj means upstream byte code.
- What flavors do we want to use for java? We already have a gcj flavor.
- What does the gcj flavor mean? Built with gcj, runs with gcj?
- Do we really need to label which compiler was used to generate byte code?
- byte code should go in a components of this package (maybe :java)
- native binary should be treated like any other native binary
