modmuss

This is a good start, however there are a few modules that have generated TAWs and/or resources. This only covers one module. Running the `generateResources` task during the github action...

> Not sure why the server test timed out [here](https://github.com/FabricMC/fabric/actions/runs/11445881667/job/31843863686?pr=4178). It doesn't seem to touch any of the code I've changed. Seems to have just been a network error when...

It doesn't improve much, I made the PR mostly so it's not forgotten about.

This has been proposed to be removed in J26: https://openjdk.org/projects/jdk/26/ Once there is an EA build for 26 this should be re tested.

Tested with the latest Java 26 EA build 17 this works fine for most older versions, but 1.4.7 is proving to be a bit tricky as it uses the applet...

Now seems to work with all versions (that I have tested), we cannot easily skip these changes pre java 26 as our own AppletLauncher class needs to compile against either...

As discussed if we make this change it should only apply to mods, and not effect the vanilla states. How we actually determine this is not super clear to me...

Do you have an idea on how to fix this? I dont think there is a solution as unfortunately there is a funimental circular dependency issue with datagen. Datagen requires...

> I have noticed that inputs also depends on arguments containing a practically guaranteed to be random field, `/tmp/loom-classpath17467358589055212817.args`. I believe this also causes caching to be practically invalid as...

This has been fixed as best as it ever will be. With newer Fabric API that datagen output is reproducable, thus if you run datagen twice it will now cache...