[DGD] importance of bytecode compatibility

[Kris Van Hees]

Just a very loose observation, but if you store your code in the snapshots,
or otherwise store it for continued use for a reasonably long time, you are
not really dealing with JIT code anymore, but rather a basic as-needed code
generation.  One of the nicer features of JIT compilation is that it is
essentially volatile... i.e. your code should continue to run perfectly fine
if e.g. you were to want to discard compiled code in order to free up some
resources.  Also, you might want to consider whether the compiled code is
something you want to be written to swap, or whether it is only really going
to make a difference for code that is used so frequently that it is not really
going to get swapped much at all.

	Kris

On Sun, Aug 07, 2011 at 05:54:19PM +0200, Felix A. Croes wrote:
> Carter Cheng <cartercheng at gmail.com> wrote:
> 
> > I am curious how important bytecode compatibility is with older
> > versions- since a very direct way of implementing a JIT on LLVM would
> > be to modify codegeni.c to generate llvm bitcode and store this
> > information instead of the bytecode in the control block data. Would
> > this be acceptable since the new vm breaks the compat anyhow?