Well purly emphatically we know nothing about compilers. Since I know what a JIT is I really like them so Im bias.
I think there has been so much work put into static compiler for lisps and other dynamic languages but there is not that much to show for it. I mean sure the lisp compilers are really fast but most of the time they get there by throwing away safty and providing typehints. While the work on self was done by one researchteam. The same with LuaJit one guy writes something for a dynamic languages and without adding typehints it starts to perform like crazy.
I think the Lisp community and the rest of the world really missed out on the JIT advantsments. Self was allready fast but Sun picked Oak (Java) to be there Weblanguage (anybody know more about that?).
WRT Self: some of Agesen's work on Self work for doing type inference on conventional, ahead-of-time compilation, and they got pretty good results this way! More interestingly, the opportunities the AOT compiler found were different than the opportunities the JIT compiler found, so there is some hope that they will combine constructively. See. e.g., http://www.cs.ucla.edu/~palsberg/paper/spe95.pdf
I think there has been so much work put into static compiler for lisps and other dynamic languages but there is not that much to show for it. I mean sure the lisp compilers are really fast but most of the time they get there by throwing away safty and providing typehints. While the work on self was done by one researchteam. The same with LuaJit one guy writes something for a dynamic languages and without adding typehints it starts to perform like crazy.
I think the Lisp community and the rest of the world really missed out on the JIT advantsments. Self was allready fast but Sun picked Oak (Java) to be there Weblanguage (anybody know more about that?).