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.
I thought about writing a JSON parser in shell, but that would be rather complex. I plan it to initialize the base for hairball so it may be used, but instead I am going to settle for copying the directories as it is far simpler.
At least for some things, I can just pull over the classes from the old class directory and save some time.
Implementing these fragments of JSR353 that I have right now is a bunch of work. Eventually when the real class libraries come I will hopefully have a something. Rather tired right now and cannot think of it. Would be a doclet. A doclet would save me tons of time slapping down all of the interfaces, methods, and stubs and such. Hopefully I can implement the entire package set in a single day, provided there are no distractions. There are only 25 classes and I have 1 stubbed and 1 fully implemented, so 23 to go. If each one takes me 30 minutes to write It will take 690 minutes or 11.5 hours. Though actually implementing the functionality on top of that would take a bit longer. It does feel like I am going in circles since I have still yet to write a compiler for Java, but each loop around gets better and better. At least now I will not have to worry about super complex dependency graphic just to build stuff with hairball, so that became far far simpler. I could add dependency graphing output for the compilation phase to determine which objects rely on which so that things could be improved or pruned. Using JSON for hairball also means I no longer need a parser for Java source code (since I used to parse tons of tags for dependency information). JSON could be used for other things also as it is more sane to edit and parse than XML. Plus, one it is fully implemented it would be the first supported JSR.
For the build system and packages I could do something similar to Debian and have a suggest or recommends kind of thing. I can make the JSON stuff I am working on now be "jsr353-spec" and have another package which implements it and provides a "jsr353-impl". That way, new implementations could be added at will and they could be auto selected. Although that might complicate things, I can just put the implementation with the jsr since there would really only be a single implementation anyway. All I know is that after all this work, I will be using JSON much.
Implementing just these interfaces is quite a task, I should avoid implementing other things that are standardized until I get something which can extract the API declarations from existing code.