What's the motivation for gcj these days?  Originally, everyone wanted a GPLd JVM so gcj kinda made sense.  At least in spirit.  It's never been a functional equivalent for an actual JVM, though.  I've seen nothing but problems with it for years in IRC channels.  It's partial implementation of the spec has led to endless confusion for uncounted newbies coming to the java channel for help.  It doesn't help that the ideologues at Debian, et. al, continue to package gcj as if it were java.  Well, we have a  GPLd JVM now.  Everything about it is open source (or just about done...). GCJ, as I see it, serves no more useful purpose than allowing those in charge of it to hold on to some ideal (or maybe pride).  I know this is inflammatory for a good number of people, but why persist?  Is it the native compilation you like?  The slow, misbegotten catastrophe that it is?  It's slower than running java bytecode and seems to eliminate several key features of Java (like dynamic classloading).  Even before Sun GPLd their (our?  I'm a Sun guy after all...) JVM, gcj adoption was miniscule at best.  So, what's the point?  Can't we move on from gcj?  Or at a minimum, stop packaging it as the default JVM when it's not actually a java implementation?  That'd work for me.  I'm just tired of seeing newbies getting tripped up by some distro's ideological navel gazing.