DISCLAIMER: These notes are from the defunct k8 project which precedes SquirrelJME. The notes for SquirrelJME start on 2016/02/26! The k8 project was effectively a Java SE 8 operating system and as such all of the notes are in the context of that scope. That project is no longer my goal as SquirrelJME is the spiritual successor to it.


Having an enumeration outside of the JavaParser class complicates things to much, as there will have to be a storage class in between. Although there cannot be per instance enumerations, the private visibility of the outer class makes things much easier to perform when setting parser options. The faux Control class would no longer be required, simple and easier is better. This will work.


Basically option parsing will go through the map of options as specified before using the private access into the class which contains what to do during compilation. In this effect, I can then unify the specifics of the code and start work on my parser. The parser would bring me one step closer to actually stepping into kernel land.


I had this idea of splitting some things off the class library into packages that would be placed in jars and put in the ext directory, rather than polluting the system class path. That would also enable the classpath to be a bit lighter than usual. This would mean splitting off everything however, but I would need a build system capable of handling such things.


Doing a major move would complicate things currently, but would be more sorted in the end. For example there is J2ME, J2SE, J2EE. J2EE extends off J2SE and has all of its packages and classes. J2ME is crippled so it is not worth doing that. I could use the ext directory to provide standard classes already split up based on the package they begin in. I can also add ones for development. The extensions may either be turned into JARs and put into the ext directory, or included directly into the system class path used by the system itself.


Otherwise without this, I would get a crowded class directory and would require more complex annotation processing (stripping the Hairball parameters) and it might be harder than usual to make sure things do not use illegal classes and packages (such as profile1 using profile3 stuff). This would also make it much more modular. I can then use package descriptions which would far better support what something is. This will slow down my compiler work for a bit, but I believe it is for the better, despite the added complexities. Conformation would be a bit easier and I could target more systems and different classes of systems. Say the Nintendo 64 could get a limited form of the libraries and such. There are 3 profiles then there is Java EE, so there will be 4 directories. Actually there are 4 profiles, the 4th being the UI stuff that is not in compact1, compact2, or compact3. When I develop the UI packages, I should not entirely depend on compact2 or compact3 so that way they can be used. I will have to figure that out when I get to it. The kernel and my own development tools could actually be their own packages also. I could also make an app directory, which is for top level applications that are not intended to be part of the class library.


I need to go into this correctly before a gigantic mess is made.


Putting everything in one single top level directory is going to be a bit messy when there are thousands of packages possibly, so I will need to categorize them.


The base root package names that I will use so far will match similar standard names within the system.


Actually splitting all of this up into packages, just solved my future problem of having architecture dependent classes, since now they can be in their own package. Also solves service handling issues since the stuff could be packed in JAR files. Also, things will now be easier to build because I do not need complex dependency grabbing, I just need tree grafting.


I believe instead of using Java files to do projects and such, I will instead use JSON, which means I will have to implement JSR353 completely and make hairball rely on it. At least JSON is easier to read than XML, and since it is in its own package, it can be included or excluded if ever needed.