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.
Have not programmed much in the past week or so. The Windows command shell is very different though in that it uses ^ for continuation on the next line (instead of \) and also & for extra commands (instead of ;). However, so far the shell stub I have written while moving the bulk of the code into Java makes it so it can be built on Windows no issues.
My double recursive building works so far, despite the JARs not getting any actual files in them.
The one good thing is I am glad I do not have to write everything else in Java 5, since Java 8 is far superior.
I can actually optimize the code so that it does not build the basic bootstrap classes twice when say compiling the kernel or running a simple package. This would greatly simplify things.
And now I no longer have double building of some of the basic JARs. Now I need to check if I can compile the stuff in Wine.
Seems that the Java compiler classes on Windows does not like the back slashes being specified as it throws an exception with an invalid relative name.
And the code reaches the same point as it does on my Debian development system, so the joys of write once run anywhere.
And Oracle Java's Swing on Windows uses DirectX, and since I am on a PowerPC system with buggy video drivers, the colors are swapped incorrectly. However, using GLX over a network connection is quite slow even if it is over a local network.
My code also compiles with GCJ and did not require any fixing, although it pops up some warnings for unused things.
What I need to do next is change the kernel building system to use the previously written about (in an earlier blog post) new way of doing the platforms with their own individual packages. The kernel builder can also be optimized to not require building a gigantic massive class library until the end so if the dynamic recompiler or the bootstrapping code fails it does not waste time building it. It would be pointless to build the class library if the dynamic recompiler cannot even recompile itself. So this would increase testing speed from a straight "-c" being passed to the launch code (which clears all of the JARs due it being a clean nature).
And with these new Json information in PackageInfo, the last remnants of the POSIX shell JSON decoder limitations has been gone away with completely.