we don't need to check this on a bit by bit level if the integers are the same.
this saves 2-3 microseconds per packet on my machine, which doesn't sound like much, but it adds up when there are lots of players.
Performance tests show that this has a considerable client-side performance advantage over using players. In my local tests, using 1000 floating texts in a 10x10x10 area, I observed an FPS increase from 1.5 to 8.0.
the following things are currently not implemented:
- particle/sound effects when an entity extinguishes itself
- particle/sound effects when mixing different stuff in a cauldron
- powder snow cauldron
both of these things are contingent on #5169, but for the time being, the PR is functionally complete and I want to move on to something else without being stalled by the particle+sound problem (which I haven't yet decided how to solve).
Bedrock block entity updates don't directly trigger block rendering
updates. This is a problem when the block entity data affects the
block's appearance directly (e.g. cauldron water colour, flower pot
contents), because it means changing them won't directly result in a
client-side render update.
This hack allows tiles to spoof block updates without actually changing
the server-side block, keeping the internals and API clean of random
shitbox workarounds.
fixes#5174fixes#4944
the old impl prevented registering more than one meta -> state mapping since the legacy numeric ID map would throw an exception if attempting to map the same ID more than once.
this was making it inconvenient for plugins to use BlockIdMetaUpgrader->addMapping(), because the block legacy ID map contains IDs up to 1.16, but the table of mapped 1.12 blockstates only goes up to ... well ... 1.12. This left a gap of several versions' blocks unable to be mapped.
