Commit Graph

1036 Commits

Author SHA1 Message Date
91d54ff0ad Removed unnecessary and/or obsolete GC hacks
the intent of these hacks was to break cyclic references to avoid having objects lingering in memory. However, all of the stuff that's being removed in this commit no longer has any effect anyway, due to the fact that these things don't circularly reference each other anymore. Notably, Tile inventories now keep Position instead of a Tile ref.
2021-06-30 19:16:25 +01:00
94e16f416d Added KnownTranslationKeys (generated) and start using it 2021-06-29 22:46:04 +01:00
32d7b1e6af Start using webmozart/pathutil for joining paths (#4287) 2021-06-29 19:40:43 +01:00
02fab77e55 World: change 'closed' to 'unloaded'
this makes more sense overall from a reader's perspective.
and also provide a rug-jerk for any idiots using World->close() when they aren't supposed to? ....
2021-06-26 21:54:18 +01:00
fc70b625b3 ExperienceOrb: Require providing XP value in constructor 2021-06-19 19:42:13 +01:00
981b0285d1 Isolate config casting nastiness in one place
this doesn't solve the underlying problem, but it does reduce the amount of noise made by PHPStan about it, as well as avoiding code litter.
2021-06-19 19:14:02 +01:00
d96fc17339 World: Check placed block collision boxes after place()
since we write these into a transaction instead of actually modifying the world directly, we can use the transaction to verify that the placement location is OK before setting the blocks.
closes #4248
2021-06-18 19:07:38 +01:00
f1583f44df LightArray: Avoid allocating 7 useless arrays for every node processed
instead, use a const array of the offsets and add them to the coordinates, which avoids the allocations.
In synthetic benchmarks, this method takes 40-50% less CPU time by eliding ZEND_INIT_ARRAY and ZEND_ADD_ARRAY opcodes. In practice, the benefit will likely be much smaller (perhaps even irrelevant).
2021-06-17 22:21:58 +01:00
71a4ea2a95 FastChunkSerializer: remove useless variable initialization 2021-06-17 20:59:40 +01:00
43f71d0d63 FormatConverter: Copy worlds for backup if rename fails
this can fail if the backups directory points to a different drive than the original worlds location. In this case, we have to copy and delete the files instead, which is much slower, but works.
I REALLY advise against putting backups on a different mount point than worlds if you plan to convert large worlds.
2021-06-17 20:46:34 +01:00
61c59be299 Replace hardcoded block metadata shifts and masks with constants
we might want to make these bigger than 4 bits in the future.
2021-06-16 12:48:09 +01:00
e7b21cf9dc World: remove obsolete TODO comment
[ci skip]
2021-05-21 13:10:48 +01:00
2ab751f985 World: fixed potential bug scheduling async light updates multiple times 2021-05-21 13:09:38 +01:00
e4750ad2cd WorldTimings: apply native typehints 2021-05-17 23:23:46 +01:00
7217ff5ff5 World: added an extra subtiming for random chunk updates 2021-05-17 23:18:56 +01:00
4736b5968d WorldTimings: change some idiotic timer names 2021-05-17 23:14:30 +01:00
cdaf734470 World: Clean up ticking chunk loader tracking 2021-05-17 23:07:48 +01:00
643c3ed14e World: Require all adjacent chunks to a ticking chunk candidate to be unlocked, generated, loaded and light-populated
this should address the remaining problems with grass spread/death.
2021-05-17 20:59:40 +01:00
7abf50f503 World: Extracted a tickChunk() method from tickChunks()
most of what's left in tickChunks() is just selecting the chunks, rather than actually ticking them.
2021-05-17 20:54:51 +01:00
bab76f4a6e World: extract some blocks of logic from tickChunks() into their own methods
the amount of nested loops, cx/dx/chunkx/etc clusterfuck in this code makes it very difficult to work with...
2021-05-17 20:50:46 +01:00
bdce781c6d Chunk: do not close preexisting tiles in addTile()
if it's desired to actually replace a tile, the old tile should be explicitly removed and closed first.
2021-05-12 12:32:36 +01:00
b2e806e2fa World: Harden chunk loading against bad data in entity/tile NBT 2021-05-12 12:26:41 +01:00
0ff21557e4 World: fixed generation requests being rejected during resolution
when a chunk population is ordered, its only chunk loader is the one that the World installed to keep the chunk loaded while it was generated. So, when the resolver removes its chunk loader from the chunk, it triggers the chunk unloading mechanism, which causes the promise to directly be rejected.
2021-05-10 17:20:04 +01:00
2ce6a87d05 getSafeSpawn() may throw if the target terrain is not generated 2021-05-08 16:22:13 +01:00
e9f3cefe94 Default world might be null 2021-05-08 16:21:34 +01:00
027109075c WorldManager: don't recalculate the spawn point for every player teleported during world unload 2021-05-08 16:19:22 +01:00
624495f4d3 Added a generic Promise type
I anticipate increasing demand for promises, and since all the libraries I could find suck, we'll stick to our own impl for now.
2021-05-08 15:57:30 +01:00
998a583815 World: Fixed inverted completion callbacks for chunk population
this would have caused any chunk populated by a plugin to remain loaded forever.
2021-05-07 23:39:44 +01:00
742f86e022 Rename DestroyBlockParticle -> BlockBreakParticle
closes #3461

literally every other particle/sound has the subject first, followed by the (optional) verb, and finally Particle (or Sound).
In addition, we refer to breaking blocks as 'break' everywhere except here, where we refer to it as 'destroy'.
2021-05-06 18:33:18 +01:00
7b2c6c5ceb Avoid more $this refs on long-life closures 2021-05-06 14:27:56 +01:00
c356abb917 BlockTransaction: remove cyclic reference to self in callbacks 2021-05-06 14:27:08 +01:00
4f4069d403 World: remove another noisy debug message 2021-04-19 00:55:42 +01:00
caf3ec1b54 World: improve documentation of World::getPotentialBlockSkyLightAt() 2021-04-18 20:31:42 +01:00
53ebe4f9f9 World: added getHighestAdjacentFullLightAt() 2021-04-18 20:20:08 +01:00
51b0685881 World: document getFullLight() and getFullLightAt() 2021-04-18 20:14:16 +01:00
f29ababf8d World: Rearrange some light-related methods
these things are all over the place ...
2021-04-18 20:07:36 +01:00
b31b097b8e LightPopulationTask no longer hard-depends on Server 2021-04-16 15:41:14 +01:00
ebd19f5a70 SubChunkExplorer: Protect world field 2021-04-15 23:32:30 +01:00
bf67170504 SubChunk: don't assume 15 sky light when not allocated
closes #2533
at long last...
2021-04-15 23:22:53 +01:00
5d83f4670a RegionLoader: Switch to using named constructors
this makes the code more self-descriptive, and also helps to detect potential bugs.
2021-04-15 21:57:23 +01:00
e8dd4de5c8 SubChunk: Lazily allocate LightArrays as needed
this is slightly slower, but saves a significant amount of memory (~80 KB per chunk).
Since ext-chunkutils2 doesn't do copy-on-write trickery like the PHP impl did, we need this to get the memory advantages back.
2021-04-15 21:43:46 +01:00
e6ecacbb2b World: do not access chunks for light if they aren't yet light-populated 2021-04-15 21:21:34 +01:00
ab9615fb9c World: fixed documentation for getHighestAdjacentPotentialBlockSkyLight()
[ci skip]
2021-04-15 21:19:58 +01:00
9655cb819d World: Remove duplicated code 2021-04-15 19:41:03 +01:00
a0c3102b05 FormatConverter: add @phpstan-param for newProvider ctor parameter 2021-04-15 19:14:52 +01:00
8e3f5737a0 RegionWorldProvider: fix CS 2021-04-15 15:39:08 +01:00
32c4a165cf Removed RegionLoader::open()
this is nothing but a source of bugs.
2021-04-15 15:38:18 +01:00
edb590f681 missed one ... 2021-04-15 15:23:15 +01:00
08f0c9a244 Separate writable parts of RegionWorldProvider into WritableRegionWorldProvider 2021-04-15 15:20:57 +01:00
e6fb6b1f27 BaseLevelProvider: Remove obsolete proxy methods 2021-04-15 15:15:16 +01:00