I am very glad to hear this. New code will always have a tendency to make the code clearer, smaller and more maintainable.
Maybe not allowing Java classes without source … … at least not in new modules.
The restriction that old modules must work in V4.
I should be very careful about commenting on VASSAL. It has worked for many years and proved itself and given many gamers many good hours. Keep up the good work.
For my own account, I start afresh with new ideas and new ways. My design is solid and has solved many issues. A guy wanted to test Laua script while the gaming was running. This has now been implemented with a Run script window. This ability to create, compile and run script while the game is running is immensely useful. Will V4 allow something like that? The scripting language is actually not a sideshow but the means by which the whole engine is driven.
Much work remains. I hope to be able to keep working on this project in the time to come.
By taking the module offline until the source is in place. Actaully it could not be possible to upload compiled classes, just the source code. The source would then be compiled and added to the module.
Last I checked, about 8% of all published modules have custom code, and only a few of those have source code inside them. We’re not likely to get the source code for many of those modules, so taking them down until we got the source code would effectively mean taking them down forever. The module library is meant to be archival, so this isn’t compatible with that aim.
It would take significant engineering work to establish a way to check that uploads contain source that matches the class files, or to compile the class files ourselves. The number of modules with custom classes but containing no source code that are yet to be uploaded has to be so small that the effort can’t possibly be justified, when the aim is to retire Java entirely.
This leads me to the question…How much of a learning curve do you anticipate there will be for those of us who are not natural born coders to build a V4 module from scratch, versus how we have been building them in V3?
I expect it to be far, far simpler with V4. Module design with V3 involves learning the quirks of a domain-specific language that evolved over 25 years in ways that make sense only in the context of its own history. We won’t need to drag any of that along for V4.