This repository was archived by the owner on Oct 31, 2023. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 47
Changelog
Guillaume Piolat edited this page Jan 20, 2016
·
28 revisions
Breaking changes:
- dropped support for compilers with < 2.067 frontend. GDC is unsupported until it reaches 2.067.
-
gfm:math
has been split in two packages. All things rather useless were put into agfm:integers
package which contains:- half floating-point numbers in
gfm.integers.half
- wide integers in
gfm.integers.wideint
- fixed-point numbers in
gfm.integers.fixedpoint
- rational numbers in
gfm.integers.rational
- half floating-point numbers in
-
gfm:image
now depends onimageformats
version 5. This shouldn't affect your code at all. - all
deprecated
aliases and functions gone. Theclose()
function are gone for real. - no more dependency on the
logger
package. From now on onlystd.experimental.logger
is used. - no more predefined aliases for matrices of unsigned integers.
Breaking changes:
-
gfm:image
now uses theimageformats
package as a dependency (https://github.com/lgvz/imageformats). The function for image loading is stillloadImage
and should work identically. This unbreaks JPEG loading.gfm.image.stb_image
is gone. No more GIF loading but I don't know if it even worked and you get TGA loading instead. -
gfm.math.funcs.min
andgfm.math.funcs.max
were removed since they already exist instd.algorithm
. -
The
close()
functions are all gone! The new behaviour of destructors is to assert if an actual resource release happen while called by the GC (this assert does not trigger in release mode). So relying on the GC for accidental resource release is now a bug. You can use.destroy
instead to call the destructor for a class object in replacement for callingclose()
. -
gfm:math
is not compatible anymore with front-ends prior 2.066
Features:
- Derelict loaders aren't unloaded anymore when the library object disappear. Should allow eg. other OpenGL libraries to work with GFM.