16:00:49pjbbeach: in the extreme case of a 1-instruction processor, you don't have to store any bit to encode the operation code. You just need to store the addresses of the arguments or the jump.
16:00:58pjbbeach: therefore code, is purely addresses.
16:02:23pjbOf course, we can color two kinds of pointers, data pointers, and code pointers.
16:03:03pjbThis is why compilers have to perform two kinds of analysis and optimization: data flow analysis and optimization, and code flow analysis and optimization.
17:35:20jackdanielebrasca: you can say exactly the same thing about the philosophy - it may be a huge waste of time and may not give you understanding. people are far too often seduced by their own words
3:24:12tichbeach: Have you started working on the LispOS?
3:25:23beachSICL is the basis of it, so you can say that, yes.
3:25:41beachThere is not much to an OS. You just have to include the bootstrapping code.
3:26:11beachIn the case of CLOSOS, there is also the checkpointing code of course.
3:27:31beachBut SICL already includes first-class global environments in the form of Clostrum, so that part is done.
3:29:41beachAlso, as I have already mentioned, a lot of the ideas of CLOSOS can be tested without it running on bare metal. It can very well be a Unix process for quite some time.
3:30:30beachI guess for "bare metal", there are also the device drivers.
3:33:56beachBut since both Mezzano and Movitz exist, there has got to be some code for doing those things already. I am more interested in the interface between applications.