effect manager

[pythoncoder1234]

• installed new library: fastutil
  • faster versions of java collections
• effect manager to manage attributes and status effects
• optimized effect managing using hashmaps and
• get/set attribute/status effect methods have been moved to effect manager
  • call movable.em().function() to access
• most of these changes were from refactoring the above, i promise there aren't that many changes

[ghostlypi]

Just an FYI, TanksJson got killed because I was trying to use GSON instead of @aehmttw's own TanksON library. I don't think @aehmttw likes adding new libs for something as small as a few percent performance boost.

[pythoncoder1234]

I think he kinda wanted to use the one he wrote in cs class... we'll see. Also the performance boost is not just a few percent, especially since the collections avoid unboxing where possible it saves a ton of memory and improves performance.

[aehmttw]

I got it to crash in arcade mode multiplayer

java.lang.NullPointerException: Cannot invoke "tanks.BiConsumer.accept(Object, Object)" because "this.addAttributeCallback" is null at tanks.effect.EffectManager.addUnduplicateAttribute(EffectManager.java:166) at tanks.effect.EffectManager.addStatusEffect(EffectManager.java:223) at tanks.effect.EffectManager.addStatusEffect(EffectManager.java:183) at tanks.AreaEffectFreeze.update(AreaEffectFreeze.java:92) at tanks.gui.screen.ScreenGame.update(ScreenGame.java:1543) at tanks.Panel.update(Panel.java:569) at tanks.GameUpdater.update(GameUpdater.java:23) at lwjglwindow.LWJGLWindow.tick(LWJGLWindow.java:399) at lwjglwindow.LWJGLWindow.loop(LWJGLWindow.java:330) at lwjglwindow.LWJGLWindow.run(LWJGLWindow.java:123) at main.Tanks.main(Tanks.java:106)

[ghostlypi]

What's the Cache Gradle action for?

[pythoncoder1234]

It's supposed to cache the gradle distribution... not 100% sure it works yet. It seems to be building faster but I'm not sure.

[ghostlypi]

I don't think gradle caching does anything. That is because github actions start each action in a new container, so we compile and cache with gradle, but never revisit the cache.

[pythoncoder1234]

my codeql builds run in less than a minute, faster than runs on the master branch here which take 2-3m. I seem to be doing something right...

[ghostlypi]

This action took 45 s to build, and it does not have caching enabled. I think that it might be a bit random tbh.

[pythoncoder1234]

it’s probably proportional to the number of classes changed then

[aehmttw]

What is the advantage of this system over the old one?

[pythoncoder1234]

• adds StatusEffect modifiers to the attributes list directly (instead of additionally having to iterate through all status effects and their modifiers)
  • better implementation in my opinion, it shouldn't make the code much more complicated
• uses more hashmaps to improve performance (mapped by name -> modifier, type -> ArrayList)
• optimized memory usage and performance using faster data structures from fastutil
  • example: iterating through attributes.values() once per tank per frame is a memory disaster, has been fixed
• split movable and effect-related stuff into different files, in a way that imo is intuitive

[ghostlypi]

I don't think your changes to the Compatibility class allow the class to do what it is designed to do. What you have done is generalized the function to convert a boolean into a TanksONable, however, you failed to consider other cases which may require more specific special cases. An example would be to take an integer value, and convert it to an object where the integer value may be a single value within the new object. Additionally, you outright remove the code that would enable different fields to be renamed and have their values carried over between game versions. This compatibility subsystem needs to be "future proof" to some degree and be able to expand over time. If you disagree with the design of a system, you should propose an alternative design first so as to not accidentally remove functionality from an existing system.

[aehmttw]

This pull request does not have anything to do with compatibility so I don't know why that's included here. If you wish to PR compatibility, please make a separate pull request. Thanks.

[pythoncoder1234]

whoops reverted it