Split-off coremod and mod classes

[Prototik]

Split-off coremod and mod classes due to classloading errors with SpongeForge.

Issue reported in #2509, first revision of PR (#2483) was rejected due to naming conventions, so I tried to keep old coremod name and give appeng.transformer.AppEng to mod class.

[Prototik]

@yueh any comments why it's closed?

[Prototik]

@yueh @shartte anyone?

[Cloudhunter]

I've been using the auto build from the PR for two days now - appears to working well. I'd much love for this to be merged into the mainline, so that I don't have to use an unofficial build just to keep Sponge!

[Prototik]

So heated discussion...

[shartte]

See #2636
If you want to contribute, you can help review that...

[KasperFranz]

@shartte but it didn't get implemented with that, and making that still a problem.

@Prototik tried to fix it with #2670

You guys are really making it hard for everyone :/, since we have to use a custom fork which includes this or similar fixes - since not running sponge is not really an option when having a large server.

[yueh]

It is still just a bandaid not a fix. Forge is even considering getting rid of coremods completely. So we might have to replace it completely. It should be doable to dynamically generate bytecode for things like TileEntity as they have a factory method. But that might not apply to every use case we need. So worst case we have to use LaunchWrapper or inject our own classloader somehow, etc.

So now is your chance to actually provide some constructive feedback about how we can remove the coremod completely, while still maintaining the same feature set. Instead of just complaining that you have to put a few minutes of effort into applying a patch and run ./gradlew build.

Because once we find and choose an approach, we will most likely stick to it as it will probably be not a trivial effort to get it working. Which can also mean it might no longer be compatible with sponge at all. Not just some random loading of forge classes (which can be trivially fixed).

At that point it will be more or less "The issue was open for a long time. You did not raise any concerns about whatever solution or suggest an approach. We consider that as tacit approval. Now deal with it."

I doubt anyone of you will appreciate a solution, which pretty much cannot be reverted anytime and completely break it for you (neither do we). But applying the band-aid now means you can ignore it and have to face the actual solution in a few weeks or months. Whatever it will be.

[Dockter]

Correct me if I am wrong but a duct tape fix is better than telling your server user base to use a custom fork or recompile it yourself isn't it?

It's my understanding that the patch the SpongeTeam recommended didn't interfere with any subsystem. Feel free to correct me if I am making the wrong assumption.