* Initial work on 1.16.100 - currently crashes the client
* Update runtime item states
* Use new Bedrock runtime IDs
Bedrock now hardcodes block runtime IDs in alphabetical order of the identifiers. This commit updates Geyser to accomodate.
- Remove runtime_block_states.dat and replace it with blockpalette.nbt
- Calculate the block runtime ID based on the order of the block palette
- Separate BlockTranslator.AIR into Bedrock and Java values
- Update the second layer of chunks to use air when not waterlogged
- Don't send item palette for now, as that's what crashes the game (will look into for v415)
- Other misc. changes
* Improve second layer chunk translation
* v415 support
- Add a message warning people they are on a beta version of Geyser
* Update to protocol v417
There are still some mappings changes that need to be gone through.
* Update runtime item states and clean up item frames
* Future-proof enchanment table
* Update for v418
* Update to v419
* Apply proper air ID to waterlogged chunk layer
* Fix missing import
* Remove beta warning
* Update mappings
* Manually patch runtime_item_states and send the ITEMS registry
* Update README
* Disable grindstone and smithing inventories (since they're broken)
* Use artifactory jenkins plugin (#1548)
* Use artifactory jenkins plugin
* Bump version to 1.2.0-SNAPSHOT
Co-authored-by: SupremeMortal <6178101+SupremeMortal@users.noreply.github.com>
* BlockStorage is never used concurrently, no need to synchronize
* initial, semi-functional, faster chunk conversion
* faster chunk conversion works well for every situation except spigot
* delete unused ChunkPosition class
* preallocate and pool chunk encoding buffers
* make it work correctly on spigot
* make field naming more consistent
* attempt to upgrade to latest MCProtocolLib
* remove debug code
* compile against my MCProtocolLib fork while i wait for my upstream PR to be accepted
* return to Steveice10 MCProtocolLib
* make ChunkPosition use a hashCode implementation with far better hash distribution
this should improve the performance when used as a hash table key
* ChunkCache no longer uses position wrapper objects
this yields a roughly 15-20% increase in performance when converting chunk data
* fix code style issues
The refactors in this commit go a bit beyond the scope of what the sound/effects branch was meant to accomplish, however most of these changes are necessary so chunk caching could be reintroduced for the standalone version. The chunk caching here allows for us to get the block ID before the block was broken, and in the future allow us to implement newer features. Chunk caching is optional (and disabled by default) as on non-Bukkit versions, it can eat up a lot of RAM with many players online.
Camotoy
DaPorkchop_
DoctorMacc
RednedEpic
Ender
SupremeMortal
AJ Ferguson