BiomeGen::getNextTransitionY(y) did not guarantee the condition (y < biome_y_min)
of the next loop because the function may return the value (biome_y_min - 1).
Hence, the biome was not updated until one Y coordinate after.
Also removes all animations but the first one from gltf_frog.gltf
to address the corresponding warning.
Catches some more possible exceptions (out of bounds, optional access)
which might be caused by a broken model to properly log them.
* Cache node in voxel area index when possible
The index function according to the MSVC profiler actually takes up a significant time slice (around ~5% of total time for the process) during normal game-play.
Might not be accurate but still good to not recalculate it twice.
* Remove `setNodeNoRef` from VM
* VM: remove old commented out print statement
`""` does not refer to a proper path, and `fs::PathStartsWith(path, "")` should just return `false`. This is also the case in libraries in other languages where I looked, seems to be common.
The new behavior:
* check early, if `prefix` is empty - return if path is empty or not,
* no special processing for when `path` is empty, the function meets characters in `prefix` and returns false anyway.
* Upgrade client active object mgr tests to Catch2
In addition to invoking Catch2's test runner after Minetest's homemade
runner, this refactors the tests to follow the DRY principle, and gives
them expressive names and clear assertions. Catch2 is already bundled
with Minetest, so there are no added dependencies.
* Increment failed modules count for Catch2 tests
* Respect --test-module option for Catch2 tests
* Improve Catch2 --test-module behavior
This switches infostream to rawstream so that test runner output is
displayed, and returns the correct boolean depending on the results. The
tests are now found by setting the configuration instead of invoking the
command line parser.
* Test uniqueness of all IDS instead of just one
Co-Authored-By: Lars Müller <appgurulars@gmx.de>
* Include Catch2 test run in timing and logging
* Flush std::cout after printing Catch results
* Increment total tests run instead of hardcoding to 1
* Flush stderr before printing to stdout
It's necessary to flush stderr before printing to stdout in adition to
flushing stdout before printing to stderr, to make sure all output is
ordered correctly.
* Make Catch write to rawstream
---------
Co-authored-by: Lars Müller <appgurulars@gmx.de>
Fixes a regression caused by 4245a760
'moveItemSomewhere' attempted to add a leftover stack to an empty stack, resulting
in an empty name with non-0 ItemStack count.
