Negative values should not disappear. For example, if _HIWORD returns 63000,
then _GET_Y_LPARAM should return -2536, but now it returns 63000.
This fixes that.
In DirectX 11, if a device and a factory are independently created,
some functions like MakeWindowAssociation doe't work well.
This change fixes the issue by getting a factory from a device and
using it.
Closes#2661
This change adds a new compiler directive 'kage:unit' to Kage. This
takes one of these two values: 'pixel' and 'texel'. The default value
is 'texel'.
With the pixel-unit mode, all the built-in functions treats pixels
instead of texels, and the texCoord argument of Fragment is in pixels.
This simplifies shader programs as programs no longer have the notion
of texels.
With the texel-unit mode, the behavior is the same as the current
behavior.
Closes#1431
This is basically a revert of 0035ba0bd1.
I couldn't confirm that allowing tearing improved FPS. Rather, this
has a potential risk to cause an error at Present of the swap chain due
to an inconsistent argument, though I have not confirmed an actual issue.
Now the DirectX 11 driver was introduced, the situation has changed.
Updates #2034
Updates #2188
This is achieved by replacing the function by one that only adds a single
new region, and only considers duplicates between the previously existing
region and the one newly added one, thereby removing previously redundant
checking of each previously existing region against each other.
This speeds up AAAAXY loading on a Moto G7 Play from 52.27 seconds to
8.15 seconds.
Closes#2626
By removing redundant work done in canAlloc and just calling alloc right away,
this removes 35% contribution to AAAAXY loading CPU time on the profile on
https://user-images.githubusercontent.com/251568/230496805-c5e32c19-9258-49c8-800b-a3f0bc3b072d.svg,
or - as measured via stopwatch - brings loading time on Moto G7 Play from
17.22s to 14.94s after already applying #2627.
This should be safe as there is no case in which alloc succeeds and the
allocated region isn't used; also, there is no case in which alloc mutates the
tree when it doesn't actually succeed (comment added in one place to justify
this).
Closes#2628
