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.


It is foggy outside. So the thing to do would be to figure a nice way to represent parts of a file which can be used by a class generator.


Thinking about it, the BDF stuff would be a bit complicated.


So it would have to be done carefully.


Ok, the BDF will be in certain sections. The uppermost sections are primary, then to secondary, and such. Or similar wording. The primary will just contain the global constant pool, references, and the set of classes. For Java class files this will just contain a single class for the most part. The primary then contains individual class information which has stuff such as flags, fields, methods, and such. So my binary format will need to have a BinPrimary and a BinClasses for the most part. For the Java class format these will just contain single entries. This way that code which permits multiple class files and such would be simplified. If another format for binaries is added then they can potentially use it.


The light reflecting off my hand is like a diffuse watch reflection.


Did some outside work since it is going to snow soon.


What I can do is have a kind of stream like descriptor thing when loading data. That would keep it semi-simple and I would be able to reverse the operation and have write support along with that.


Yes I believe a stream-like approach would be best. When reading and writing files I can just do the whole thing at once, although it will be a bit slower there would be no need for a cache. However, I could still support caching for reading since it would be faster to not load everything at once if it is never needed at all.


This refactor will need the KernelBuilder to be partly rewritten, but that stuff needs to be split anyway because the OS builder is a bit messy.


Probably about time I do that planned refactor of redoing the configuration system for the OS. New plan is to have CPUs and machine in JSONs with potential "base-off" fields. This cuts into my plan of having an actual running kernel by the end of the month. However belaying this will cause a much larger mess in the future. However there is code that is already refactored such as the standard RISC translators so that does not need to be touched at all. And the Bin* classes are fine also.


This new configuration system should be simpler to use and less messy when it comes to things for example. I can also support stuff such as searching for target hardware based on the manufacturer or the name of a board, perhaps with regular expressions and such.


With a PackageContents class having MergeJar is no longer needed to build the OS since the contents would be handled in a map of sorts. This can also handle the odd case where some files are the same between JARs. MergeJar would just replace entries which might end up confusing things. If I keep up the programming work though, I should get back to where I was before if I keep at it all day. However, distractions are tough.