The global(!) table settlements_in_world` has no use, but is serialized and
saved to a file `settlements.txt` every time during generation of a new
village, adding more lag. This commit removes all related code:
* Remove all instances of `settlements_in_world`.
* Remove `settlements.load()` and `settlements.save()`.
* Move descriptive function comments to above the function definition.
* Adopt some luadoc styles.
* Update a stale reference to a renamed function in comments.
* Fix silly and mostly inconsequential logic bug that had gone unnoticed up
till now.
* Condense a single use variable away by using a slightly more elaborate
assignment statement.
* Add a few more local aliases for global minetest.* functions.
* Replace use of local table with compostability values with a call to
`minetest.get_item_group()`.
* Define local alias for `minetest.get_item_group`.
* Remove the now unused static compostabilty values table.
mcl2 uses add_particle for nether dust resulting in a 10-fold
increase in network traffic when in the nether. Nether dust is not
configurable making it impossible to turn this off for server admins.
this commit replaces the add_particle method with particle spawners
* For some unexplained reason, `mcl_burning.storage[player]` can
sometimes be `nil`, causing a crash in `on_leaveplayer()`. This
commit adds a check for that. If a `nil` is encountered, a sane
value is substituted and a warning is set to the log.
* Change carpet from `group:attached_node` to `group:supported_node`,
allowing carpets to be placed on top of torches, water source blocks
and other non-walkable nodes, like in MC.
* Add support for `group:supported_node` to CORE/mcl_attached.
Supported nodes are nodes that can be placed on any node that does not
have the `drawtype = "airlike"` attribute.
* Copy the `drop_attached_node()` function from minetest/builtin, so
that the override function provides the same behavior when nodes drop.
* Add comments to CORE/mcl_attached and to the functions defined in it.
* Add more local aliases for global minetest.* functions.
* If the original function returns true, it is not necessary anymore
to perform more tests and the override function can simply return true
immediately.
* Remove unused code that was commented out. The code tried to find
fire luaentities in the same spot as the newly created fire luaentity.
It may have been intended to optimize getting set on fire multiple
times, but it makes no sense as it does not discriminate between fire
luaentities attached to the object set on fire and those attached to
other objects. The function that this code was in also has a better
way to prevent adding multiple fire luaentities in the first place.
* In mineclone5 a crash was reported to occur when deserialization of
storage returned nil in on_joinplayer. https://git.minetest.land/
MineClone5/MineClone5/commit/96c4fb60d8641b4181edb902ed24dbf173828d09
This commit uses a different, but equally effective fix.
* Remove animation_frame from fire entity state, it is now kept in the
storage table of the parent entity.
* Rename animation_timer in fire entity table to _mcl_animation_timer,
in line with mineclone2 policy on adding custom members to minetest
tables.
* Comment out code that does nothing sensible. Scheduled for deletion
at a later time.
* Make carpets walkable. When a carpet is placed on top of a fence or a
wall, the carpet can be jumped upon, allowing to in effect jump over
fences and walls.
The daylight detector code used get_node_light(), which detects both
natural and artificial light. This patch improves the code to use
get_natural_light(), which was introduced with Minetest 5.4.0.
* Check if node has a definition table before attempting to evaluate its
attributes. By application of De Morgan's law, the compound logic
expression can be rewritten so that the existence of the table is
checked before all following attribute references.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Clarify code flow.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Mobs redo uses `registered_items[]` where clearly the proper table
accessed should be `registered_nodes[]`. Perhaps this magically
works, but it looks nonsensical. Switch to the use the more
sensible `registered_nodes[]` table.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Check if node has a definition table before attempting to evaluate its
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Check if node has a definition table before attempting to evaluate
attributes.
* Define local variable to cache multiple accesses to `registered_nodes[]`
and improve readability.
* Reduce redundant `== false` condition check.
This solves a problem were nether portal removal would trigger deep
recursion through node callbacks. For large portals this could result
in stack overflow crashes on LuaJIT. The issue is solved by rewriting
the portal removal to avoid recursion and removing the portal in one
operation using minetest.bulk_set_node.
Changes to mcl_burning.set_on_fire():
* Add logic that only updates the burn time, but skips adding fire
entities to objects that are already burning.
* Condense code a little, remove single use variables.
* Add a comment to a questionable piece of code.
* Add comments to the function.
* Add update_hud() function.
* Remove the client channels.
* Unglobalize animation_frames variable.
* Fix bug where player state was not stored on shutdown of singleplayer
game.
* Remove superfluous sanity_check() function, this code could easily be
inlined in its only caller.
The parameters for the flames sprite were incorrect, causing the reverse
side of the sprite to appear as an opaque black rectangle. Use the
correct incantation.
Before this patch, the tga_encoder mod would write corrupted TGA files
on Windows: Bytes that looked like newlines were replaced by a carriage
return and a newline.
During the rewrite in 89e55e9065, a queue system was added to allow
using both VoxelManip-based generation steps, and set_node-based steps,
however some set_node-based steps (underground mushrooms, nether
decorations, and structures) were missed and remained in a VoxelManip
step. The result is that the changes made by the set_node stages were
overwritten after the (now-stale) VoxelManip data was committed later.
(ref. Mineclonia/Mineclonia#26)
The function dynamic_add_media() was changed in incompatible ways in
several minor versions of Minetest, breaking the display of handheld
maps in Minetest 5.5.0. This patch makes handheld maps display there.
The function was blocking with one argument in Minetest 5.3. It was also
blocking in Minetest 5.4, but took an additional argument for a function
to execute once the media had been received. Calling dynamic_add_media()
with a single argument had been deprecated; a function that did nothing
was provided in mcl_maps to satisfy the changed argument requirements.
In Minetest 5.5, dynamic_add_media() was changed to non-blocking. This
introduced a race condition in mcl_maps, where a client often tried to
display a map before it had received the map texture from the server.
Opening an issue on the Minetest issue tracker led to it being closed in
about 20 minutes: <https://github.com/minetest/minetest/issues/11997>
Redstone comparators have two modes, comparison mode & subtraction mode.
Before this patch, the functions to turn comparators on or off attempted
to swap nodes with comparators in the same mode, but failed to determine
the correct replacement node, if the existing node was not a comparator.
When a comparator in an on state (e.g. powered by a filled cauldron) was
flooded, the flooding dropped the comparator and replaced the comparator
node that was to be swapped out with air, which lead to a server crash.
This patch changes the functions that turn comparators on or off so they
only swap existing nodes with comparators in the same mode if the name
of the replacement node can be determined – i.e. if it is not nil.
In commit 55009c257e that added vectors to
mcl_composters, mcl_hoppers was accidentally patched with a unrelated
change, updating it to get_item_group(). This mostly works, but in one
particular case the semantics of the return value differs. Instead of
returning 'nil' it returns '0'. That altered the evaluation of an if
condition, breaking the abm that sucks in items.
This commit fixes the conditional by explicitly comparing '~= 0'.
The function on_place_bucket defined a couple of variables that were
used only once or in a redundant check. After removal of the redundant
check in a previous commit, all use-once variables can now be substituted
with their assignment expressions.
The function on_place_bucket contains the following logic:
if not a and not b then return x end
if a then
foo(a)
elseif b then
foo(b)
else
return x
end
The "if not a and not b then .." is removed because the case is handled
by the else case later on. This will allow some further simplifications.
Instead of directly accessing the group table in node definitions, use
the proper minetest.get_item_group accessor to test for "cauldron"
group.
Also adds local alias get_item_group for the global minetest call.
Two instances of add_node and add_item still used the full minetest
call, substitute the alias.
Added a set_node alias for several minetest.set_node calls.
When using a bucket of water on a cauldron, this would not fill the
cauldron as expected, but deposit a water source block on top of the
cauldron instead.
Applied patch from mineclone5 #38 (commit #
698c29733f06a7fcb7e755bf26ee46b33b00699b) from mineclone5 that fixes
this problem.
Implements a composter block with crafting recipe, and some more
uncraftable blocks representing various stages of filling.
Adds a list of items that can be used with the composter and the chances
of these items adding a layer to the composter.
Implements methods to add compostable items to the composter, to update
the composter to various levels of compost and to harvest an item of
bone meal when the composter is ready.
Textures were taken from XSSheep/Nova_Wostra Pixel Perfection texture
pack.
In Minetest 5.4.1, calling get_player_control() on a mob returned the
empty string. Minetest commit 5eb45e1ea03c6104f007efec6dd9c351f310193d
changed this, so now calling get_player_control() on a mob returns nil.
As mcl_boats defines boats that can have a player or a mob as a driver,
code like the following crashes with a changed get_player_control() API:
local ctrl = driver:get_player_control()
if ctrl.sneak then
detach_object(driver, true)
end
Furthermore, once a world has crashed, joining it near a mob that is the
driver of a boat with such control code immediately crashes again.
When I reported this bug to Minetest, several Minetest core developers
stated that they disliked the old API and proposed other return values
for calling a mob's get_player_control() function – all different from
the empty string. Since I have some doubts that this bug will be fixed
in Minetest 5.5.0, boat code must take into account a nil return value.
Minetest issue: https://github.com/minetest/minetest/issues/11989
- enchanted loot generated by mapgen now uses PseudoRandom for randomness
- prevent fishing loot from generating loot 32767 times (!!!) when only 1 is needed
- bows and fishing rods obtained from the treasure section of fishing loot are now enchanted
- there is now a function to uniform enchant items other than books
5.2 is actually half of the estimated MC creeper defuse range, which is 10.4.
The reason for this change is to balance the creeper in MCL2 where it fuses
whilst moving making it more difficult than MC. In MC, the creeper does not move
while fusing.
The crash occurs if mcl_burning:data is deserialized to nil. The cause of
mcl_burning being set to "return nil" is unknown. Therefore, when it occurs, it
will be logged as warning.
Source: https://minecraft.fandom.com/wiki/Tint
Since the MineClone2 biomes do not entirely match with the Minecraft ones I picked the Minecraft biomes that seem to match them best.
This also changes the palette index of the nether to match the desert instead of the mesa biome
and changes the color of grass blocks in item form to the default minecraft one.
- remove enable_bed_night_skip setting (mcl_playersSleepingPercentage setting can be used to enable disable night skip)
- make it possible to change mcl_playersSleepingPercentage ingame
- fix weird bug which allowed only numbers <= 0
- Use upright_sprite for 3rd person flame display, which is both closer to minecraft and allows for client side texture animation
- Take care of flame HUD in the MineClone2Client
Some items, like shulkers or books, can have so much metadata that the
corresponding item entity can not be serialized by the Minetest engine.
Without this patch, dropping such an item and then moving away crashes
Minetest, as it can not serialize the entity with serializeString16()
when unloading a map block.
The patch resets the overlong metadata of non-serializable item entities.
This avoids a crash and makes it possible to retrieve a “sanitized” item
without metadata when the mapblock containing the item entity is reloaded.
Originally sfan5 guessed the maximum possible item entity serialization size
that would not lead to a crash as 65530 bytes, but anon5 calculated it to be
actually 65487 bytes. This has been experimentally verified by erlehmann.
Before this change, about 10 to 30 FOV packets were sent from the server
to each connected client each second. This patch only sends FOV packets
when the FOV actually needs to be changed, i.e. when the player starts
or stops sprinting.
Several mods set or unset the visibility of a HUD bar way too often (e.g.
in a globalstep handler), causing the server to send a lot of superfluous
HUDCHANGE packets to each client. Returning from hb.hide_hudbar() early
if HUD bar visibility would not change prevents sending these packets.
