10:26
Okay so I am going to do idea simplification and combination. My idea now consists of doing the simple pool-only with statics references for the bootstrap pool initialization the same as it is done in SummerCoat's initialization sequence. But it will have it be stored in the ROM instead as an initial memory boot sequence. It will effectively be both but it will instead just involve static memory and a sequence of instructions used to initialize it, either based off the RAM offset or JAR offset. So it will end up being much simpler but a combination of the two. There is no need to have an insanely complex boot initialization process when I have proven that just the constant pool being initialized with pointers is enough. This initialization could be done by the bootstrap as well.
16:54
It is too complex to have to multiply by 4 to get the class pointer.
17:23
So I simplified VTables for special and instance invokes. The only thing for me to do are interface invocations. These are a bit complex because I will have to do some lookup to figure out what is what. So the lookup will be done much and very often, so how is it made fast and rather constant? How about if there were a known interface index. Basically what would happen is that... well it might not work because the object might not be known at compile time and it would be complicated otherwise. I think it would be best to have a helper method instead of having this sit in the code compiler somewhere. But it should have three arguments: the class of the object being invoked, the interface to be invoked, and the method index of the interface method being invoked. I think that is the simplest way.