this reduces the footprint of RuntimeBlockMapping by a further 1 MB, as well as simplifying various parts of the code, and solidifying the immutability guarantee of BlockStateData.
This commit completely revamps the way that blocks are represented in memory at runtime.
Instead of being represented by legacy Mojang block IDs and metadata, which are dated, limited and unchangeable, we now use custom PM block IDs, which are generated from VanillaBlocks.
This means we have full control of how they are assigned, which opens the doors to finally addressing inconsistencies like glazed terracotta, stripped logs handling, etc.
To represent state, BlockDataReader and BlockDataWriter have been introduced, and are used by blocks with state information to pack said information into a binary form that can be stored on a chunk at runtime.
Conceptually it's pretty similar to legacy metadata, but the actual format shares no resemblance whatsoever to legacy metadata, and is fully controlled by PM.
This means that the 'state data' may change in serialization format at any time, so it should **NOT** be stored on disk or in a config.
In the future, this will be improved using more auto-generated code and attributes, instead of hand-baked decodeState() and encodeState(). For now, this opens the gateway to a significant expansion of features.
It's not ideal, but it's a big step forwards.
this commit provides a central place where all block data can go to be upgraded to the latest version (currently 1.19), irrespective of how old it is.
Previously I had issues during debugging, because it wasn't possible to just upgrade a block without deserializing it into a Block object, which isn't currently supported for many blocks.
This commit solves that problem by separating the upgrading from the deserialization.
I've stuck to only doing this in the places where I'm sure we should never get false back. Other places I'm less sure of (and I found more bugs along the way).
this fixes many cases of corruption during disk-full situations - file_put_contents() would write an empty file, destroying the original data.
fixes#3152
this is not as good as phpstan/phpstan-src#769 (e.g. array_key_first()/array_key_last() aren't covered by this, nor is array_rand()) but it does eliminate the most infuriating cases where this usually crops up.
This reverts commit a5418a019dc2a83210084632130db8ce06f529ea.
The more I assessed this, the more I realized that this implementation
doesn't actually offer any value. Since modcounters don't persist after
chunk unload + reload, they can't be reliably used to detect changes in
chunks without additional event subscriptions.
For the purpose I actually intended to use them for (population task
cancellation) there's a) another solution, and b) modcounts are
unreliable for that too, because of the aforementioned potential for
chunks to get unloaded and reloaded.
For the case of detecting dirty chunks within PopulationTask itself,
they are also unnecessary, since the dirty flags are sufficient within
there, since FastChunkSerializer doesn't copy dirty flags.
In conclusion, this was a misbegotten addition with little real value,
but does impact performance in hot paths.
