This is so, but why are our language A <-> language B translation tools so bad, that people end up having to manually convert code from one language to another (e.g. if some code exists in language A, but you need its functionality in language B)?
because some of the essence of what you wanted to express is lost when you compile. it's not a one to one mapping.
otherwise we would just compile and decompile into each form. intermediates like lvm come close to trying to do this i think? not sure if they allow going back to the original form from the IL though.
oh, and lastly, think of trying to express a feature in 1 high level language with another that doesn't support that feature. it's quite hard to gain expression when it's not idiomatic in that language.
kind of similar to spoken language actually. some concepts in some languages form as single words or phrases that never really translate well to other spoken languages because the place that other language formed never had that concept.
Compilers have gotten better, but in comparison to a human, they are still not particularly good at the edge cases. Things like instruction selection and register allocation in particular, sometimes I wish I had the time to write my own more intelligent algorithms for them.
