These antialiasing techniques have been removed in OpenGL 3.1, they were often executed by the CPU, and Minetest does not use them.
The OpenGL wiki recommends that we do not use this functionality in our program.
https://www.khronos.org/opengl/wiki/Multisampling#Smooth_antialiasing
1. we don't need it
2. it's dropped in SDL 3
3. it breaks compilation on Alpine and postmarketOS with SDL2 enabled
(and the bug never going to get fixed on SDL2 side)
Signed-off-by: David Heidelberg <david@ixit.cz>
Minetest using RTTI, so we cannot apply the flag here
if we want to start using IPO/LTO.
If we keep `-fno-rtti` here, compiler loses part of informations and
cannot do efficient optimizations.
The benefit of using RTTI on Irrlicht disappears with IPO/LTO on whole binary.
Signed-off-by: David Heidelberg <david@ixit.cz>
The size of a mip map is `max{floor(width / 2 ^ level), 1} x max{floor(height / 2 ^ level), 1}`,
where `width x height` is the size of the full-resolution image,
`level` is the integer mip map level
and the smallest mip map has `1 x 1` resolution.
If `regenerateMipMapLevels` is called with custom mip map data,
the mip map sizes are calculated in this function and separately in `uploadTexture`.
`uploadTexture` calculates a size by `floor(width / 2 ^ level) x floor(height / 2 ^ level)`.
To support non-square textures, after this change,
`uploadTexture` sets the mip map width or height to `1` if it is `0`.
it's enabled inconditionally on android
also remove some _IRR_COMPILE_WITH_ANDROID_DEVICE_ which are useless because cmake is already properly configured
Image size calculation could overflow s32 in one place (but not others where it was done correct), which first lead to wrong amount of memory getting allocated for image data and later crash in the CColorConverter.
Thanks @sfan5 for his fuzzing tests @https://github.com/minetest/irrlicht/issues/236
and @erlehmann for passing them on: https://irrlicht.sourceforge.io/forum/viewtopic.php?t=52925
Also updating changes.txt with TGA loader changes from this and previous commits.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6535 dfc29bdd-3216-0410-991c-e03cc46cb475
From sfan5's fuzzing test reported in Minetest here: https://github.com/minetest/irrlicht/issues/236
Was missing test if it writes beyond allocated memory which can be triggered by TGA's which lie in their RLE data.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6534 dfc29bdd-3216-0410-991c-e03cc46cb475
Was creating 16-bit images for those before.
Could also support 24-bit images, but either we need another convert function or another palette for that (the 16 and 32 bit both work with 32 bit palettes, the 24 bit conversion function only with 24 bit palettes)
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6533 dfc29bdd-3216-0410-991c-e03cc46cb475
TGA's can claim to use less palette colors than they later do.
We only support 8-bit palettes, so to make this safer lets just always allocate at least 256 bytes.
Thanks @erlehmann for report and testcase: https://irrlicht.sourceforge.io/forum/viewtopic.php?p=307191
Based on Minetest bug report: https://github.com/minetest/irrlicht/issues/236
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6532 dfc29bdd-3216-0410-991c-e03cc46cb475
Note: Due to OSX always failing merge OSX it's rather applying a second patch
from Ryan Schmidt from bugreport #462 to trunk
(Not sure why svn merging always fails for OSX. Probably related to MacOSX files getting moved in the past. Was that done without svn move commands? I guess that can't be fixed anymore now)
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6521 dfc29bdd-3216-0410-991c-e03cc46cb475
This also fixes the combobox which users the listbox like that.
This got broken in [r6454] which had fixed the events send out by the listbox
Note: Still a bit strange behaviour when leaving the combobox at the bottom, but that was already that way in Irrlicht 1.8, so I've got to investigate that on it's own.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6463 dfc29bdd-3216-0410-991c-e03cc46cb475
We had changed that once before in the other direction in svn r421
Reason back then was "Sleep(0) doesn't allow any lower priority threads to execute"
But Microsoft changed the behaviour of Sleep(0) after Windows XP so that's no longer true.
And the costs of it is pretty high - due to this using a timer with a 15ms resolutions it meant not just giving up the thread but it also always waited for 15ms on Windows.
I also replaced a few sleep calls in examples for that reason with yield() calls.
git-svn-id: svn://svn.code.sf.net/p/irrlicht/code/trunk@6459 dfc29bdd-3216-0410-991c-e03cc46cb475
