A new month is today.


I made 731 commits last month, but this total should drop down in the future because my note taking system is no longer part of the repository. I mostly just worked and refactored the JIT for the most part. I am slowly getting to the point where I will have native code. Hopefully this month I get to that point. One other thing I did was that I refactored the build system to be in SquirrelJME itself so that all of the related build code is internal. If I write a Java compiler then SquirrelJME would be truly self hosting and could very much build itself (provided there is enough memory and disk space to do so). Most of the code I wrote earlier this month has survived as there are really no issues with it at all. I added some new classes such as the multi-set Deque which is useful for register allocation where the same set of registers are in multiple queues. I also implemented a sorted map and set. But most of the work was mostly JIT, which is slow going but getting there. The JIT is the most important part of the code because it defines the future for the most part, so one must be very careful. I added the base for the LUI interfaces and the system used to implement display drivers. I added a bunch of utility scripts that would make things easier for me. I had the generic JIT but that has gone away for the basic JIT, but I am slowly reworking that JIT to be a bit better with this new executable system that I plan to use.


The executable system would hopefully be a way where I can keep code and data alone for the most part. I made some ZIP code fixes. The newly created build system is much nicer though and has removed a bunch of duplicate code.


I also constantly refactored the JIT much.


One thing is that my commit count should become more accurate rather than a bit inflated due to the note taking changes.


Removing all of those notes caused my file count to take a dip however. My line count also changed too.


Thinking about it, I can have the most recent test results stored in the unversioned space also. However, one issue with that is that markdown has no concept of tables or colorings. However, I can use emphasis and strong for fail/pass. I could always generate CSV or RTF, or perhaps find a text based rich formatting that is mostly just text data.


One thing is that the doclet API does not really have parent class handling. Basically everything is dot form as it appears in source code compared to the form used in class files.


One script that will be complex is the one that syncs changes into the unversioned space in the repository.