Paper 1.21.11 builds do not support Spigot NMS mappings anymore?

[EricLangezaal]

Expected behavior

Dear all,

I have developed the PetDragon plugin(s), which are used by customers on Spigot and Paper alike. These plugins rely extensively on NMS (internal Minecraft code).

I develop my plugin against Mojang's mappings, but then use the SpecialSource Maven plugin to remap my plugins to be Spigot compatible. For years, this also meant they were PaperMC compatible.

However, with version 1.21.11, an updated plugin that correctly finds NMS classes on a SpigotMC server fails to find those classes on a PaperMC server, instead giving NoClassDefFoundErrors.

If I build my plugin jar without the SpecialSource Maven plugin, it finds most classes on Paper except the v1_21_R7/Craft.... classes.

Since I would like to support both Paper and Spigot customers, I sincerely hope PaperMC is planning on making a 1.21.11 version that actually supports Spigot-based NMS plugins, like for all previous versions.

Thank you in advance,

Observed/Actual behavior

With version 1.21.11, an updated plugin that correctly finds NMS classes on a SpigotMC server fails to find those classes on a PaperMC server, instead giving NoClassDefFoundErrors.

If I build my plugin jar without the SpecialSource Maven plugin, it finds most classes on Paper except the v1_21_R7/Craft.... classes.

Steps/models to reproduce

Make a dummy plugin, have it important at least one NMS class, and then remap it using SpecialSource to be Spigot compatible. This plugin will now function on Spigot-1.21.11, and not on PaperMC for 1.21.11.

Plugin and Datapack List

Paper version

This server is running Paper version 1.21.11-51-main@90191f7 (2025-12-21T22:07:10Z) (Implementing API version 1.21.11-R0.1-SNAPSHOT)

Other

No response

[electronicboy]

The mappings are coming, just needs validating manually which takes time at a busy part of the year; I will also note that Mojang are also dropping their obfuscation in the next release, so you should probably look towards supporting mojang mappings soon™

[mbax]

We hard forked from Spigot over a year ago. Plugin devs have had over a year to support Paper, which has the vast majority of users. And we've made it remarkably easy with good documentation and support.

I suggest using paperweight-userdev to best accomplish this. As someone who uses it regularly, I find it much easier to use than other options.

Track #13388 if you still need time to migrate your tooling.

[EricLangezaal]

The mappings are coming, just needs validating manually which takes time at a busy part of the year; I will also note that Mojang are also dropping their obfuscation in the next release, so you should probably look towards supporting mojang mappings soon™

Ah perfect, glad to hear they are coming soon! Any chance you could ping me here when they do (or like where I can see that?:) Usually they have long been implemented by the time users ask me to update PetDragon haha -- so was curious why it didn't work suddenly.

And yes will definitely support them once Spigot does too! Which I think they are planning to do by 2026.1 basically.

[EricLangezaal]

We hard forked from Spigot over a year ago. Plugin devs have had over a year to support Paper, which has the vast majority of users. And we've made it remarkably easy with good documentation and support.

I suggest using paperweight-userdev to best accomplish this. As someone who uses it regularly, I find it much easier to use than other options.

Track #13388 if you still need time to migrate your tooling.

This is not relevant. Firstly, Spigot's plugin community and number of plugins/downloads is multitudes bigger. I therefore develop for Spigot, since that usually makes plugins compatible anyway. Most importantly, this (mostly) concerns a Premium (paid) plugin -- for which I cannot host a version on Paper's Hangar since it lacks that functionality entirely.

[mbax]

If you still need to support Spigot due to not wanting to move to another platform (there are others), it's still fairly trivial to build in support for both.

You could have had all but one of your users (I checked your bstats - literally one Spigot server) able to run on day one, by being able to dev and test in advance because we were pushing out snapshot/pre-release builds. Instead you're stuck waiting almost 2 weeks post-release so far for support by targeting Spigot exclusively. Just a thing to consider.