This implement branch prediction macros likely() and unlikely() like in
Linux. When using these macros please ensure that when you use them the
condition in the branch really is very, very likely or unlikely. Think
90+% of the time. Primarily useful for error checking. (And I implement
these macros for all our error checking macros now)
See this article for more information:
https://kernelnewbies.org/FAQ/LikelyUnlikely
There are more places where these macros may make sense in renderer and
physics engine. Placing them will come in another commit down the line.
operator= does not need to call reference() if the new value is of the
same type as the old. This saves us zeroing the Variant, This speeds
up reuse of a Variant in a loop by roughly 50%.
Previously godot_variant_new_object constructed Variant without
accounting for the fact that the Object can be a Reference, so refcount
was not increased and References were destructed prematurely.
Also, Reference::init_ref did not propagate refcount increment to the
script instance, which led to desync of refcount info on the script
side and Godot side.
Currently we rely on some undefined behavior when Object->cast_to() gets
called with a Null pointer. This used to work fine with GCC < 6 but
newer versions of GCC remove all codepaths in which the this pointer is
Null. However, the non-static cast_to() was supposed to be null safe.
This patch makes cast_to() Null safe and removes the now redundant Null
checks where they existed.
It is explained in this article: https://www.viva64.com/en/b/0226/
There was a logic error in #7815 which made
Variant.hash_compare() == Variant.hash_compare() always true.
In an attempt to short-circuit the NaN check I made an (in hindsight) obvious
error: 10 == 12 || is_nan(10) == is_nan(12)
This will be true for all inputs, except for the NaN, not-NaN case. The macro
has been updated to now generate:
(10 == 12) || (is_nan(10) && is_nan(10))
so:
(10 == 12) || (is_nan(10) && is_nan(12)) = false
False or (False and False) is False
(10 == 10) || (is_nan(10) && is_nan(10)) = true
True or (False and False) is True
(Nan == 10) || (is_nan(NaN) && is_nan(10)) = false
False or (True and False) is False
(Nan == Nan) || (is_nan(NaN) && is_nan(NaN)) = true
False or (True and True) is True
Which is correct for all cases.
This bug was triggered because the hash function for floating point numbers
can very easily generate collisions for the tested Vector3(). I've also added
an extra hashing step to the float hash function to make this less likely to
occur.
This fixes#8081 and probably many more random weirdness.
I can show you the code
Pretty, with proper whitespace
Tell me, coder, now when did
You last write readable code?
I can open your eyes
Make you see your bad indent
Force you to respect the style
The core devs agreed upon
A whole new world
A new fantastic code format
A de facto standard
With some sugar
Enforced with clang-format
A whole new world
A dazzling style we all dreamed of
And when we read it through
It's crystal clear
That now we're in a whole new world of code
- Add FIXME tags comments to some unfixed potential bugs
- Remove some checks (always false: unsigned never < 0)
- Fix some if statements based on reviews.
- Bunch of missing `else` statements
This fixes HashMap where a key or part of a key is a floating point
number. To fix this the following has been done:
* HashMap now takes an extra template argument Comparator. This class
gets used to compare keys. The default Comperator now works correctly
for common types and floating point numbets.
* Variant implements ::hash_compare() now. This function implements
nan-safe comparison for all types with components that contain floating
point numbers.
* Variant now has a VariantComparator which uses Variant::hash_compare()
safely compare floating point components of variant's types.
* The hash functions for floating point numbers will now normalize NaN
values so that all floating point numbers that are NaN hash to the same
value.
C++ module writers that want to use HashMap internally in their modules
can now also safeguard against this crash by defining their on
Comperator class that safely compares their types.
GDScript users, or writers of modules that don't use HashMap internally
in their modules don't need to do anything.
This fixes#7354 and fixes#6947.
Made sure files in core/ and tools/ have a proper Godot license header
when written by us. Also renamed aabb.{cpp,h} and object_type_db.{cpp,h}
to rect3.{cpp,h} and class_db.{cpp,h} respectively.
Also added a proper header to core/io/base64.{c,h} after clarifying
the licensing with the original author (public domain).
That year should bring the long-awaited OpenGL ES 3.0 compatible renderer
with state-of-the-art rendering techniques tuned to work as low as middle
end handheld devices - without compromising with the possibilities given
for higher end desktop games of course. Great times ahead for the Godot
community and the gamers that will play our games!
A general speedup should be apparent, with even more peformance increase when compiling optimized.
WARNING: Tested and it seems to work, but if something breaks, please report.
