Performance profiling on Linux AMD64 showed this to be a significant bottleneck. The non-inlined functions are expensive due to XMM registers spilling onto the stack.
Sometimes you need to be able to do removal-related cleanup, such as removing files from disk, or entries from a database. staticdata obviously isn't suitable for large data. The data shouldn't be removed if the entity is unloaded, only if it is removed.
New users find Minetest's account system confusing.
This change moves username/password to a new dialog,
with login and register buttons added to the Join Game tab.
The old registration confirmation dialog is removed in
favour of the new dialog.
Fixes#8138
* fix integer overflow in mapgen
Some calculations involving the magic seed had overflow because the result of an intermediate arithmetic step could not fit in an s32. By making the magic seed unsigned, the other operand in the equation will be cast to unsigned, and possibly other operands or intermediate operands. This will result in unexpected behavior if an operand is negative, which is technically possible, but logically should not happen.
* comment noise2d bitshift
While working through the code I was momentarily concerned that the right bitshift in noise2d could fill ones in some cases. It turns out that with signed integers, this is indeed true, but this one is shifting an unsigned integer, so the behavior is as expected. I put a comment here to clarify this, in case someone else wonders the same thing down the line.
* noise2d and noise3d unittests
I have added 3 tests each for noise2d and noise3d, testing all zero inputs, a very large seed (case which caused UB in the old implementation) and some fun primes I picked for no particular reason. This should be sufficient to demonstrate that the behavior of the new implementation has not changed. I used uniform initialization because it is a good feature of C++11. Please do not explode.
* uncomment the noise2d bitshift
This reverts commit 583b77ee9f1ad6bb77340ebb5ba51eb9a88ff51c. It's a
well-defined language semantic; it doesn't need to be commented.
* code cleanliness
This was added a long time ago in 42bbd5c9ae06a8d8ffb7915599097ead6f848755
and meant to fix prevent the view becoming black when jumping into a
ceiling, this no longer happens today.
On my system this is a reduction from 4664 to 3704 bytes.
This is not for the sake of saving RAM but ensuring
commonly used structures fit into caches better.
stringstreams were shown to be slow when reading/writing single characters
and there is lots of potential by having functions perform on existing
buffers whenever possible.
* Rewrite shadow filtering for the new distortion
* Calculate penumbra radius using a single sample
* Avoid peter-panning effect due to filtering of short shadows
* Add adaptive filter quality for soft shadows
* Avoid sharp shadows on surfaces without normals (e.g. plants)
* Increase default and maximum soft shadow radius
* Make line numbers in shader errors match the code
This reduces the
(absolute) time spent in Server::SendBlocks() from 700ms to 300ms
(relative) share of MapBlock::serialize() from 80% to 60%
in a test setup with 10 players and many block changes
The logging streams now do almost no work when there is no output target for them.
For example, if LL_VERBOSE has no output targets, then `verbosestream << x` will return a StreamProxy with a null target. Any further `<<` operations applied to it will do nothing.
If a newly started thread immediately exits then m_running would
immediately be set to false again and the caller would be stuck
waiting for m_running to become true forever.
Since a mutex for synchronizing startup already exists we can
simply move the while loop into it.
see also: #5134 which introduced m_start_finished_mutex
This commit introduces mt_localtime() in src/gettime.h, a wrapper
around the OS-specific thread-safe versions of localtime()
(resp. localtime_s on Windows and localtime_r in other systems).
Per the Open Group recommendation,
«portable applications should call tzset() explicitly before using
ctime_r() or localtime_r() because setting timezone information is
optional for those functions», so we also do a one-shot
call of tzset() (_tzset() on Windows to avoid warning C4996).
The function is used to replace the localtime() calls in
getTimestamp() and makeScreenshot().
(The only reminaing call to localtime() in the tree now is the one in
the local copy of the Lua source code.)
As reported in #12197, b0b9732359d43325c8bd820abeb8417353365a0c
introduces a regression in worldalign textures.
The specific change that seems to be responsible for this issue is the
change in order between the computation of the cuboid texture
coordinates and the box edge correction.
Fix#12197 by moving the box edge correction back to before the cuboid
texture coordinates, as it used to be.
Since b2eb44afc50976dc0954c868977b5829f3ff8a19, a texture defined as
`[combine:16x512:0,0=some_file.png;etc`
will not be sent correctly from a 5.5 server to a 5.4 client due to the
overeager detection of unsupported base modifier `[` introducing a
spurious `blank.png^` before the modifier.
Fix this by whitelisting which base modifiers can be passed through
unchanged to the client, and prefix `blank.png` for the others
(which at the moment is just [png:, but the list may grow larger
as new base modifiers are added.)
Stop scaling images to POT immediately when loaded. The 'combine'
modifier hardcodes X and Y coordinates, and so behaves incorrectly
if applied to a scaled image. Images emitted by generateImage()
are already scaled to POT before being used as a texture, so
nothing should break.
* Move shadow position calculation to vertex shaders
* Animate entire scene before rendering shadows to prevent lagging of shadows
* Remove unnecessary use of PolygonOffsetFactor
* Apply normal offset to both nodes and objects
* Rename getPerspectiveFactor -> applyPerspectiveDistortion
* Remove perspective distortion from fragment shaders
* Pass perspective distortion parameters as uniforms
* Set all perspective bias parameters via ShadowRenderer
* Recalibrate perspective distortion and shadow range to render less shadow geometry with the same quality and observed shadow distance
Change 1: Clarify when on_step collision information is provided
Change 2: Document PostgreSQL and Redis settings
Change 3: Overall AreaStore documentation improvements including consistent parameter naming based on community suggestions
* Allow to set the displayed item count and its offset via meta
* fix rect constr call
* devtest: add dump_item chatcommand
* fix rect2 constr call (sdim is a position (typedef for v2s32), not a dimension) and remove background because it would work now
* add missing utf8 to wide conversion
* rename to count_meta
The dynamic shadows are yet not in the desired state to justify the
inclusion into version 5.5.0. A stable release is long overdue, hence
this allows fixes to continue in 5.6.0-dev to finally release an
acceptable version of the dynamic shadows feature.
Reverting this commit is highly recommended to proceed in development.
fixes:
* Switching between games does not immediately hide creative mode / damage buttons if so specified
* World creation menu has a game selection list even though the menu already provides a gamebar
* Showing gameid in world list is unnecessary
* Choice of mapgen parameters in menu persists between games (and was half-broken)
Fix HTTPFetch caller and request ID to 64 bits
Check that allocated caller ID is not DISCARD
Print body if serverlist request returns error
Don't print control characters from HTTP responses
Document special HTTPFetch caller IDs
Allow unicode to be printed
This fixes a nullptr dereference when the specified inventory list is not known.
Happens when HUD elements are sent before the required inventory list is created.
It's likely that these could be used trick mods into revealing the insecure
environment even if they do everything right (which is already hard enough).
This ensures that no overflows (side-effects) happen within the find_nodes_in_area function by limiting coordinates like done in the map generation code.
The length check used < rather than <=, disabling the check when the formspec version
matches the client's FORMSPEC_API_VERSION.
Additionally, it was possible to have fewer parts than required if the formspec version
was greater than the client's FORMSPEC_API_VERSION.
From November 2021, the Play Store will no longer be accepting
apps which use the deprecated getExternalStorageDirectory() API.
Therefore, this commit replaces uses of deprecated API with the new
scoped API (`getExternalFilesDir()` and `getExternalCacheDir()`).
It also provides a temporary migration to move user data from the
shared external directory to new storage.
Fixes#2097, #11417 and #11118
Code that relies on `resend_count` was added in 7ea4a03 and 247a1eb, but never worked.
This was fixed in #11607 which caused the problem to surface.
Hence undo the first commit entirely and change the logic of the second.
- iconv() prototype changed from traditional Unix defintion
to POSIX definition in 9.99.x.
- wchar_t is not a valid character set for iconv. Share code
with Android for using UTF-32.
This commit deprecates the forward, backward, left, and right binary
inputs currently used for player movement in the PlayerControl struct.
In their place, it adds the movement_speed and movement_direction
values, which represents the player movement is a polar coordinate
system.
movement_speed is a scalar from 0.0 to 1.0. movement_direction is
an angle from 0 to +-Pi:
FWD
0
_
LFT / \ RGT
-Pi/2 | | +Pi/2
\_/
+-Pi
BCK
Boolean movement bits will still be set for server telegrams and
Lua script invocations to provide full backward compatibility.
When generating these values from an analog input, a direction is
considered active when it is 22.5 degrees away from either
orthogonal axis.
Co-authored-by: Markus Koch <markus@notsyncing.net>
Co-authored-by: sfan5 <sfan5@live.de>
Some games provide users with enough freedom to create items
with metadata longer than 64KB, preventing this from causing
issues is on them but we'll still do the minimum not to abort
the server if this happens.
When setting a sprite animation, do not keep the last animation's frame number. Setting a new animation should start the animation at the start of the new animation.
* The problem were lines like these:
"# textdomain:"
* str_split does not add an empty last part if there is a delimiter
at the end, but this was probably assumed here.
* Add a simple PNG image encoder with Lua API
Add ColorSpec to RGBA converter
Make a safety wrapper for the encoder
Create devtest examples
Co-authored-by: hecktest <>
Co-authored-by: sfan5 <sfan5@live.de>
With the CMake changes to IrrlichtMt, it's now possible to use a target for IrrlichtMt.
Besides greatly improving the ease of setting up IrrlichtMt for users building the client, it removes the need for Minetest's CMake to include transitive dependencies such as image libraries, cleaning it up a tiny bit. The PR works by finding the IrrlichtMt package and linking to the target it provides. If the package isn't found and it isn't building the client, it will still fall back to using just the headers of old Irrlicht or IrrlichtMt.
Reduces stutter and freezes when playing.
* Maintains double SM and SM Color textures
* Light frustum update triggers incremental generation of shadow map into secondary 'future' textures.
* Every incremental update renders a portion of the shadow draw list (split equally).
* After defined number of frames (currently, 4), 'future' and 'current' textures are swapped, and DirectionalLight 'commits' the new frustum to use when rendering shadows on screen.
Co-authored-by: sfan5 <sfan5@live.de>
Apparently the C++ standard library is supposed to provide
specializations of std::hash for enums (even in C++11)
but those don't always work for whatever reason.
* Separate draw list and shadows update to reduce jitter
* Avoid draw list update and shadow update in the same frame
* Force-update shadows when camera offset changes
* Adjust minimum filter radius for perspective
* Expand shadow frustum when camera FOV changes, reuse FOV distance adjustment from numeric.cpp
* Read shadow_soft_radius setting as float
* Use adaptive filter radius to accomodate for PSM distortion
* Adjust filter radius for texture resolution
when minetest.remove_detached_inventory is called on script init, the environment is yet not set up,
hence m_env is still nullptr until all scripts are loaded
Add API for mods to hook liquid transformation events
Without this API, there is no reliable way for mods to be
notified when liquid transform modifies nodes and mods are
forced to poll for changes. This allows mods to detect
changes to flowing liquid nodes and liquid renewal using
event-driven logic.
This completely removes any mention of the software and D3D drivers from MT, preventing the user from accidentally attempting to use them. Users who need a software renderer should be asked to install Mesa drivers which offer superior fidelity and performance over the 'burningsvideo' driver.
Fixes list clearing for inv:set_list() using same size, since 2db6b07.
addList() now clears the list in all cases. Use setSize() to resize without clearing.
If enabled in minetest.conf, provides colored, clickable (middle-mouse or ctrl-left-mouse) weblinks in chat output, to open the OS' default web browser.
This permits to make evidence that we have some bad object passing on various code parts. I fixed majority of them to reduce the scope of passed objects
Unfortunately, for some edge cases i should have to expose ISceneManager from client, this should be fixed in the future when our POO will be cleaner client side (we have a mix of rendering and processing in majority of the client objects, it works but it's not clean)
* we don't need on CAO side more than SceneManager, and temporary. Pass only required SceneManager as a parameter to build CAO and add them to the current scene
* Use temporary the RenderingEngine singleton from ClientEnvironment, waitfor for better solution
* Make ClientActiveObject::addToScene virtual function mandatory to be defined by children to ensure we don't forget to properly define it
* No more access to the singleton instance from everywhere (RenderingEngine::get_instance dropped)
* RenderingEngine::get_timer_time is now non static
* RenderingEngine::draw_menu_scene is now non static
* RenderingEngine::draw_scene is now non static
* RenderingEngine::{initialize,finalize} are now non static
* RenderingEngine::run is now non static
* RenderingEngine::getWindowSize now have a static helper. It was mandatory to hide the global get_instance access
Make the RenderingEngine filesystem member non accessible from everywhere
This permits also to determine that some lua code has directly a logic to extract zip file. Move this logic inside client, it's not the lua stack role to perform a such complex operation
Found also another irrlicht <1.8 compat code to remove
