DOSBox-Staging Git (2023/06/24) is compiled. This repository attempts to modernize the DOSBox project by using current development practices and tools, fixing issues, adding features that better support today's systems, and sending patches upstream.

DOSBox-Staging Git Changelog:
* Add unit tests for the RGB conversion functions
* Create capture output directory lazily
* Simplify format_string() helper function
* Clean up manual memory management when capturing images
* Make libpng a hard dependency so image capturing is always available
* Add file_open() wrapper over fopen() to pacify MSVC
* Add reminder about chunking audio captures into 2GB WAV files
* Address PVS Studio & various other CI warnings
* Move image capture classes into src/capture/image
* Factor out ImageCapturer from capture.cpp
* Always pass Fraction by reference
* Update English translation file
* Remove unnecessary include
* Assert bits_per_pixel value is valid
* Update VS Studio project files
* Fully implement new image capture overhaul
* Change the capture file naming scheme to be less surprising and to
* Propagate the type of the screenshot to the image capturer
* Rewrite capture filename generation to use std_fs::path
* Add CapturedImageType enum
* Use std::string in CAPTURE_CreateFile
* Add format_string() helper function
* Add strip_prefix & strip_suffix string util functions
* Remove SDL2_image
* Make capturing rendered (post-shader) images threaded (SDL texture)
* Only capture the contents of the OpenGL viewport
* Make capturing rendered (post-shader) images threaded (OpenGL only)
* Add support for vertically flipped images to the image capturer
* Add explanatory comments to capture functions
* Always pass RenderedImage by reference
* Switch to using fractional pixel aspect ratio in the image capturer
* Switch to using fractional pixel aspect ratio in render.cpp|h
* Extract pixel size calculation into a function & clean it up
* Remove old float-based pheight and pwidth calculation and keep the new
* Remove the old one_per_pixel_aspect calculation and keep the fraction
* Use fractional pixel aspect ratios in vga_draw.cpp
* Rename RenderedImage_t to RenderedImage
* Add class to work with fractional numbers
* Rename capture_image.cpp|h to image_capturer.cpp|h
* Make image capture threaded
* Log error if the PNG library cannot be initialised
* Improve capture log message
* Document PNG library calls & add explanatory comments
* Improve PNG compression parameters for best speed & compression ratio
* Introduce RenderedImage_t for passing image related parameters sets around
* Normalise the order & names of the the render & capture image variables
* Retire capture flags
* Rename dblw & dblh flags in render.cpp to double_width & double_height
* Implement & use new image upscaler for capturing images
* Rename the very misleadingly named "pixel aspect ratio" variables
* Rename SDLMAIN_GetRenderedSurface to GFX_GetRenderedSurface
* Remove ancient aspect ratio fudging/hack
* Add functions to convert between linear float RGB and 8-bit sRGB
* Pass pixel aspect ratio to capture image
* Use RGB helper classes and change PNG pixel format to RGB (from BGR)
* Add extra constructor and ToRgb888() methods
* Use less ambiguous rgb666 name instead of rgb18
* Add Rgb555
* Use RGB LUTs in Rgb565
* Add fast "magic" RGB 8-bit to 5/6-bit conversion functions
* Switch to faster "magic" RGB 5/6-bit to 8-bit conversion functions
* Add 5-bit to 8-bit RGB LUT support to rgb.h
* Extract 6-bit to 8-bit RGB LUT to rgb.h
* Use vector instead of a fixed-length array for row_buffer
* Remove unnecessary/error-prone host_to_le from Rgb888
* Simplify 24-bit data handling
* Rename Rgb16 to the more descriptive Rgb565 name
* Rename rgb24 to the more descriptive Rgb888 name
* Enforce packing for packed "rgb24" (RGB888) pixel values
* Simplify image capture to write raw rendered image without width/height doubling
* Fix capture video width & height handling
* Fix capture image width & height handling
* Break apart capture_image into separate functions
* Add description about the list of valid `bpp` (bits-per-pixel) values
* Improve mode table formatting
* Don't set vga.draw.double_scan flag as it's not read anywhere
* Fix license header
* Disable writing command line history if entry is empty
* Fix MCB Fault Strategy config option description
* Reorder '[dos]' config section content
* Cleanup a Coverity null-deference reloading the shader
* Ignore warnings from GCC 9 CI workflow
* Update and consolidate Linux CI jobs
* Update and consolidate macOS CI jobs
* Add latest clang and gcc native files for Meson
* Ignore new MSYS2 tar warnings during extraction
* Fix an unused variable when logging in PCI
* Cleanup a couple uninitialized array warnings in Voodoo
* Replace custom UINT64 with standard type in Voodoo
* Replace custom INT64 with standard type in Voodoo
* Replace custom INT32 with standard type in Voodoo
* Replace custom UINT32 with standard type in Voodoo
* Replace custom INT16 with standard type in Voodoo
* Replace custom UINT16 with standard type in Voodoo
* Replace custom INT8 with standard type in Voodoo
* Replace custom UINT8 with standard type in Voodoo
* Standardize prefix when logging in PCI
* Ensure the previous video mode uses a randomly-access iterator
* Ensure CacheBlock uses a standard object layout
* Standardize the CacheBlock with struct types

Download:
DOSBox-Staging Git (2023/06/24) x64 : 1cloudfile bayfiles gofile multiup pixeldrain send
Source: Here