In order to make CommandQueueMT more maintainable this PR changes the
previous macro hell with variadic templates instead. This makes the
class far more explicit and will allow us to more easily change the way
the class functions in the future.
Furthermore this refactoring has allowed for some optimizations. In
particular by using std::forward to delay the decision of decaying the
type to as late as possible we are able to move the data from the
callsite into our Command buffer and later move it to the call.
In practice what this means is that compared to the old version instead
of copying values 3 times, we can now get away with 1 copy, and 1 move
for lvalues, and just 2 moves for rvalues. This saves quite a few
operations in a hot codepath.
We also now test to make sure that the amount of copies and moves are
what we expect. This way we can spot performance regressions in this
code easily.
Somewhat unscientifically, running TPS-demo by pressing enter and not
touching the controls average mspf, repeatable across many runs:
before: 6.467
after : 6.202
This fixes UBSAN errors reported by running our testsuite, importing the
TPS demo, and running the TPS demo. I have tried, wherever possible, to
fix issues related to reported issues but not directly reported by UBSAN
because thse code paths just happened to not have been exercised in
these cases.
These fixes apply only to errors reported, and caused by, core/
The following things have been changed:
* Make sure there are no implicit sign changing casts in core.
* Explicitly type enums that are part of a public API such that users of
the API cannot pass in wrongly-sized values leading to potential stack
corruption.
* Ensure that memcpy is never called with invalid or null pointers as
this is undefined behavior, and when the engine is built with
optimizations turned on leads to memory corruption and hard to debug
crashes.
* Replace enum values only used as static values with constexpr static
const values instead. This has no runtime overhead. This makes it so
that the size of the enums is explicit.
* Make sure that nan and inf is handled consistently in String.
* Implement a _to_int template to ensure that all of the paths use the
same algorhithm, and correct the negative integer case.
* Changed the way the json serializer precision work, and added tests to
verify the new behavior. The behavior doesn't quite match master in
particulary for negative doubles as the original code tried to cast -inf
to an int. This then led to negative doubles losing all but one of
their decimal points when serializing. Behavior in GDScript remains
unchanged.
Some tests have been removed since there's no way to guarantee they will
pass.
Other tests have been refactored to ensure proper waiting, and taking
into account potential out-of-order delivery (which is unlikely in test
scenarios but expecting a specific order on a UDP socket is wrong and
OSes makes no promises of ordered delivery on localhost).
"Should disconnect client" test was failing randomly on Mac CI tests, so
this PR is making it more reliable reading on the closed client instead of
writing to it
Comparing locales can have surprising outcomes since it standardizes
locales with defaults. For example, zh and zh_CN result in an exact
match since the defaults change them both to zh_Hans_CN. Expose the
add_default parameter publicly with a default of false so the fully
standardized locale can be inspected.
Since 9456454109, country specific locales are always preferred over
country-less locales even when the OS locale country doesn't match. For
example, running the Godot editor with locale es_ES will result in the
es_AR locale being chosen even though the es locale would be better.
The change happened because the score of the es_AR locale and the es
locale are the same when comparing to es_ES. Change this by parsing
locale strings into a Locale structure and decreasing the score when
script, country or variant are set in both but not matched. For the
es_ES case, this causes the es_AR score to be decreased since the
countries don't match. On the other hand, the es locale is not decreased
since it doesn't specify a country.
Closesgodotengine/godot-proposals#5983
Adds put/get methods to `StreamPeer` that handles half precision
floating point values.
Adds endode/decode half precision floating point to `marshalls`.
Adds `get_half` and `store_half` to `FileAccess`
Co-Authored-By: "Alfonso J. Ramos" <theraot@gmail.com>
This helps, for importers spitting out new resources to the res://
filesystem to actually hash them to generate deterministic UIDs.
This PR also fixes the determinism for translations.
- Ensure String::num_int64, uint64 returns an empty string for bases less than 2 or greater than 36.
- Added corresponding test cases to verify the behavior.
- Error messages are printed when invalid bases are encountered. These messages are suppressed in the test output.
Implements support for all engine types in JSON encoding/decoding
Co-Authored-By: Juan <reduzio@gmail.com>
Co-Authored-By: Rémi Verschelde <rverschelde@gmail.com>
This is useful for custom tagging of objects with properties (for example in Blender) and having this available in the editor for scripting.
- Adds import logic to propagate the parsed GLTF extras all the way to the resulting Node->meta
- Adds export logic to save Godot Object meta into GLTF extras
- Supports `nodes`, `meshes` and `materials` (in GLTF sense of the words)
This switches to 64-bit integers in select locations of the Image
class, so that image resolutions of 16384×16384 (used by
lightmap texture arrays) can be used properly. Values that are larger
should also work.
VRAM compression is also supported, although most VRAM-compressed
formats are limited to individual slices of 16384×16384. WebP
is limited to 16383×16383 due to format limitations.
* Separates find_intersects from test_intersects for rays, and wraps the former.
* Changes parameter name to "r_intersection_point".
* Fixes broken old version which returned per axis t.
* Returns whether the ray origin is within the AABB.
* Returns intersection point when origin outside.
* Returns "backtracking" intersection point when inside.
* Returns sensible normal when inside.
* Returns valid results on borders.
* Returns robust results dealing with floating point error.
Co-authored-by: Claire Blackshaw <evilkimau@gmail.com>
* Replaces `find(...) != -1` with `contains` for `String`
* Replaces `find(...) == -1` with `!contains` for `String`
* Replaces `find(...) != -1` with `has` for containers
* Replaces `find(...) == -1` with `!has` for containers
Random-access access to `List` when iterating is `O(n^2)` (`O(n)` when
accessing a single element)
* Removed subscript operator, in favor of a more explicit `get`
* Added conversion from `Iterator` to `ConstIterator`
* Remade existing operations into other solutions when applicable
* Disabled 2D navigation tests as they do not work
* Disabled tests for primitive meshes
* Unbound some `Mesh` methods that rely on 3D resources
* Unexposed `World3D` as it depends on physics (and isn't useful)
* Unexposed some `Viewport` vrs related properties that had unexposed
methods
This adds a new enum `KeyLocation` and associated property
`InputEventKey.location`, which indicates the left/right location of key
events which may come from one of two physical keys, eg. Shift, Ctrl.
It also adds simulation of missing Shift KEYUP events for Windows.
When multiple Shifts are held down at the same time, Windows natively
only sends a KEYUP for the last one to be released.