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.
Time to do the refactor. First thing to do I say would be to cache the binary and source contents based on dates.
Ok, so using a cache for PackageContents did not take very long.
I can also probably change how PackageInfo is done and lazily cache that too because package information might only be used once before it is redone. The information on the actual information stored in the packages does not need to be in memory at all times and with the build process is only used a few times.
However, that would have to wait until after PackageContents is refactored because I can then use that instead.
I can also remove the hardcoded requirement of forcing the source to always be in a directory, so that now it can be in a ZIP so to speak, at least for getting package contents. If it is a file then it is treated as a ZIP, otherwise it is treated as a directory.
I can also make PackageInfo cached so that the JSON data does not have to
be loaded at all times. Then after this I suppose I will streamline KBF
writing. Then once KBFs are written- well I suppose I should make the CPU
and machine configurations cached and then move onto removing the need for
ArchitectureData with services and instead just have it all in a
base architecture data package where the packages using it (for example
PowerPC) will just contain the manifests which I already have. I would however
have to dump the Java stuff, however I must prepare to eat now.
And I ate. So I will continue refactoring. Now I will make PackageInfo cached. However since PackageStatus uses it, I will need a kind of bind or similar on the package contents.
Then using these changes I can support source code in the source path in the form of ZIP files.
This will also reduce duplicate code and such.
This also removes the need for binary and source directories, however the source directory may be read only so cannot remove that. However searching can be done in both locations.
The code looks much better and easier to read too.
And now that refactor is done, just going to build every package to get to
where I am at to make sure it is correct and it works. Then I have to check it
in Wine to make sure it works also. These new changes are much better because
there is potential for less memory to be used now because the
now on demand. The unification of having sources and binaries in the same
potential locations also makes things easier to use.
And it works.
I actually need reference counting for the package contents so that I do not accidently close it while it is being read since the contents are shared between multiple threads potentially.
So now to move into streamlined KBF file output, then after that removing the gigantic instruction enumerations so only databases are used instead.
AutoCloseable on the KBF to write data and then finish with the table
of contents at the end can work out nicely. The table of contents must always
be written last so this can help ensure that it is actually written.
Well my refactor still works in Wine.