My preference order of what we bundle into the .kpar is:
- Leaving the .json files unchanged, besides minimal changes such as updating a field (checksum etc) without changing structure
- Reformatting them in a canonical and "pretty" way for human readability
- Reformatting it to a space saving single line kind of thing
To have these files be readable is an upside (even from within a .kpar)
- it can help curious people learn and verify their understanding etc.
To have them mostly identical to the original version is also an upside
- if we have a error when validating a .kpar bundled .json file reporting a certain line number - then that line number is more accurate than if we made significant changes.
- it avoids introducing unexpected behavior, which raises questions
To save data space is worth very little in comparison, especially since the file will be part of a .kpar (zip) that can combine misc space characters etc to save the kind of space we could trim.
My preference order of what we bundle into the .kpar is:
To have these files be readable is an upside (even from within a .kpar)
To have them mostly identical to the original version is also an upside
To save data space is worth very little in comparison, especially since the file will be part of a .kpar (zip) that can combine misc space characters etc to save the kind of space we could trim.