So I got the Amiga documentation books that I need to port SquirrelJME to the Amiga so this is definitely going to be a reality.
Okay so for parsing classes, name resolution will be the most complex and it will require some rather complex handling to determine what everything refers to and such. There will definitely be some recursive elements to the parses and the big thing will be determining what a name means.
Actually to determine some base things, I think it would be best to scan the entire class path for classes and determine which classes and packages are available to the compiler. Also have an idea, the package info could be processed in any order for the most case. Also the compilation unit rather needs a name too.
Basically, the compilation unit parser could check, well maybe the syntax parser could perform some basic checks to make sure things are sane.
Actually, I do not need to worry about some parts of the syntax such as those in the blocks of code for methods, just need to know them for fields and annotations for the most part. I am just building a structure of things such as classes and really just need to perform resolutions and such as to what refers to what.