Skip to content

Instantly share code, notes, and snippets.

@Patbox
Last active March 24, 2024 05:11
Show Gist options
  • Save Patbox/e44844294c358b614d347d369b0fc3bf to your computer and use it in GitHub Desktop.
Save Patbox/e44844294c358b614d347d369b0fc3bf to your computer and use it in GitHub Desktop.
Why you shouldn't use cardboard, banner or any other "hybrid" server

By looking that you are here you most likely use Cardboard/Banner/some other weird bukkit + Fabric project (or just someone linked to it).

Don't do it, it will cause more harm/instabilities to your server than good. But let's start at the beginning.

Few definitions before we start:

  • Bukkit - It means bukkit, spigot, paper and any forks of these,
  • Cardboard - mod by IsaiahPatton that tries to port bukkit ecosystem to fabric.
  • Banner - another mod trying to make plugins work with Fabric, similar to Cardboard.
  • Fabric - a Minecraft modding platform, includes Fabric Loader and Fabric API, everything written in this gist also applies to Quilt.
  • Mods - Mods designed for fabric and compatible platforms
  • Plugins - Mods/plugins designed for bukkit/spigot/paper or their forks.

Why bukkit + Fabric will never work well

Mods being mods, they can modify game in substantial way. New blocks, items, mobs, etc. Possibilities are more or less endless as long as there is good enough programmer behind it. However, most mods still are kept compatible with each other, as they use tools designed for that (including Minecraft's internals, which allow for quite good compatibility) and having many gameplay mods is a norm.

Bukkit plugins on the other hand, while also normally compatible with each other, aren't really designed for custom blocks/items/etc. There are plugins that try to emulate mods with custom items/blocks/etc (for example Slimefun or ItemAdder), but they are mostly facade over vanilla items with some additional nbt. Because of that there are no issues most of the time (until other plugins don't try to use these items that is, as they see them as vanilla ones).

However, when you start mixing bukkit with fabric, bad things can easily happen. Fabric is close to Minecraft internal code (which has lots of nice and powerful stuff), while Bukkit is mostly an API which most of it's parts and design not being updated since 1.7. This means, you might need to make lots of conversions (with some not really possible without losing some functionality, for example messages in bukkit are just strings using legacy formatting, but mods use vanilla's more advanced json text for formatting). And if you start including things like non vanilla items, blocks and entities, it starts to break even more. Bukkit based server software is also known to break datapacks, which makes it even more likely to not work with mods (as a lot of them use and expend datapack funtionality).

Why Cardboard won't work

Cardboards approach is more or less taking bukkit code and hitting it with a hammer until it starts to barely works. It uses lots of hacks, code overriding entire methods in Minecraft (breaks lots of mods mixing into them!) and breaking vanilla behaviour. It's also questionable license wise, as GPL isn't really compatible with Minecraft code (and for this reason spigot and others use workarounds to bypass it). Additionally, it needs to modify some plugin on runtime to make them work, which only shows how bukkit wasn't designed for that usage. This mod will never be in a state stable enough to work with more advanced plugins while also not causing other mods to break or straight up crash. Most of the time developers of incompatible mods and plugins won't support it anyway as these mixed environments were always hacky and problematic, so it's safer and stabler to just use fabric alternatives.

Why Mohist/Banner won't work

In similar way of cardboard, to support bukkit api and plugins using it, it needs to make a lot of invalive changes to make things work. Which will also break a lot of mods (and plugins). It also is known to modify/swap plugins to make them work, so it's not even doing it's job properly (see https://essentialsx.net/do-not-use-mohist.html)

Why bukkit based software shouldn't really be around anymore

This is more of my own opinion (through it's shared by many), that bukkit based plugins and servers should be thing of the past. Bukkit (or Spigot and Paper) never fully adapted to modern Minecraft, as their goal of forever backwards compatibility limited how they could evolve. Bukkit was originally created for early Minecraft Betas and it's api was kept mostly the same. It was extended with new functionality, but it never fully adapted to modern versions. Lack of nbt api's, lack of support for datapacks as parts of plugins (which both non-bukkit platforms have supported for long time), it limits what could be done with server mods. Nearly all plugins that do something more advanced, require usage of non-bukkit Vanilla classes, either via another library or own code (and a lot of reflections and jvm hacks). For bigger servers that want to extend gameplay in their own ways, it's more common to just fork Bukkit/Spigot/Paper instead and add things in same ways as modders do.

OK, what should I do?

You should check out this list of server side fabric/quilt mods. There is lots of alternatives to pick, with even more being actively created!

If you are create mods or plugins you are also free to ask me about anything related, as I'm always happy to help people expand Fabirc/Quilt server side ecosystem :)

You should also checkout these server side communities:

Or modded ones in general:

@Patbox
Copy link
Author

Patbox commented Jul 30, 2023

I just don't want people to come to my discord/github and complain that things aren't working because of banner.

Yeah, the LOGGER.error("You won't get any support as long as it's present!"); achieves that

As with mohist, banner improves over time and solves issues, I dont think adding this to all your mods is fair play here, but that was it, sorry for bothering

You didn't read the first part? It won't work properly, so the first part is also true, there is no reason to change it. I'm not going to soften it because someone from likes hybrids doesn't like the message. I'm more tired by this entire cardboard-banner lineage breaking stuff than anything

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment