Set window icon when the emulator is started

[blackknight36]

This window icon is based on the image displayed on the web site. Fixes #90

[mist64]

Works on Mac, but

• the file needs to be taken from the directory the executable resides in. main.c already has logic for this.
• does SDL support loading a PNG instead of a BMP?
• asterisk with the variable name, not the type, to be consistent with the existing code :P

[FreeFull]

Should the icon maybe be embedded inside the binary, rather than being a separate file?

[blackknight36]

@FreeFull That is possible I just thought it would be easier to store the data in a file. I'll work on making the changes requested tomorrow.

[blackknight36]

It appears that the SDL base library only support bitmaps however there is the SDL_image library which supports other image formats.

http://gigi.nullneuron.net/gigilabs/loading-images-in-sdl2-with-sdl_image for example.

@mist64 I'm not sure if including the image path matters since the data gets compiled into the binary.

[madebr]

My 2c:

• Create a script that generates a header/source file from a file (python?)
• Add a build rule to the build system
• Compile and link the generated source
• Use SDL_LoadBMP_RW to use the file.

[blackknight36]

@madebr I've actually been working on a lua script to export bitmap data as hex values however I haven't quite been able to make it work yet. It's a lot tougher dealing with binary data than I thought. 😄

[madebr]

@blackknight36 #64 is adding CMake support to enable building on other platforms such as Windows and finally adding CI for all platforms.
Python is standard available on all these platforms.
I've taken a try at it and code is at this gist.
Use it as ./embed_c.py icon.bmp -o icon.c -p icon.h -n icondata

[blackknight36]

@madebr Thanks, I'll give that a shot. I've almost got my own script working as well, just need to fix a few bugs when dealing with padded bitmap data.

[mist64]

Sorry I haven't commented on this for a while. As I understand the code, it doesn't do any error checking if the file is not found, right?

[blackknight36]

@mist64 That is correct. I can add error checking if needed.

[blackknight36]

@madebr I've updated this PR to use SDL_LoadBMP_RW based on the data created by the embed_c.py script. I had to remove the "const" definition to get compilation to succeed. With the const keyword present I get errors as follows.

4:15 $ make
cc -std=c99 -O3 -Wall -Werror -g -I/usr/include/SDL2 -D_REENTRANT -Iextern/include -Iextern/src -c cpu/fake6502.c -o cpu/fake6502.o
cc -std=c99 -O3 -Wall -Werror -g -I/usr/include/SDL2 -D_REENTRANT -Iextern/include -Iextern/src -c memory.c -o memory.o
cc -std=c99 -O3 -Wall -Werror -g -I/usr/include/SDL2 -D_REENTRANT -Iextern/include -Iextern/src -c disasm.c -o disasm.o
cc -std=c99 -O3 -Wall -Werror -g -I/usr/include/SDL2 -D_REENTRANT -Iextern/include -Iextern/src -c video.c -o video.o
video.c: In function ‘video_init’:
video.c:164:2: error: passing argument 1 of ‘SDL_RWFromMem’ discards ‘const’ qualifier from pointer target type [-Werror]
  SDL_RWops *rw = SDL_RWFromMem(icondata, sizeof(icondata));
  ^
In file included from /usr/include/SDL2/SDL_audio.h:36:0,
                 from /usr/include/SDL2/SDL.h:36,
                 from video.h:11,
                 from video.c:9:
/usr/include/SDL2/SDL_rwops.h:164:36: note: expected ‘void *’ but argument is of type ‘const char *’
 extern DECLSPEC SDL_RWops *SDLCALL SDL_RWFromMem(void *mem, int size);
                                    ^
cc1: all warnings being treated as errors
make: *** [video.o] Error 1

[madebr]

Additionally,
I would put the icon.xcf and icon.bmp in an assets folder.

[blackknight36]

I've removed the icon.xcf file which now lets the build pass.

[mist64]

Thanks for your work, and sorry this is taking so long. I don't want to have a dependency on Python for the build process, so can you check in the resulting .c/.h files in addition to the script and remove the dependency in the Makefile (just make it an optional target)?

[blackknight36]

@mist64 No problem, I've updated this branch as suggested.

[mist64]

CI is unhappy because of spaces vs. tabs in icon.c

[blackknight36]

CI is unhappy because of spaces vs. tabs in icon.c

I've updated the script to use tabs instead of spaces. CI checks should work now.

[blackknight36]

This is no longer needed.