Separation of header and c files ?

[ejahandar]

Hi, i can't figure out why nanovg_gl.h contains both definitions and declarations together, if i want to use nvgCreateGLES2 on multiple files so i have to include nanovg_gl.h (with -DNANOVG_GLES2_IMPLEMENTATION) in each files, now the problem occurs, there are so many files which have nvgCreateGLES2 compiled in and they're conflicting with each other in link time. For the moment, i have separated definition and declarations in two separate file, am i correct?

[kritzikratzi]

i think the idea is that you only have the implementation in one file, ie get rid of the -DNANOVG_GLES2_IMPLEMENTATION, then pick one cpp file where you do:

#define NANOVG_GLES2_IMPLEMENTATION
#include "nanovg_gl.h"

every else you just include the header. (or... something like that :) )

[ejahandar]

I agree with one file for implementation, but combining both header and source file will make problems if you need using nvgCreateGLES2 on multiple files say multiple surfaces.

[memononen]

The one file idea is taken from stb libraries [1]. You should have only one file in your project where you define the implementation as @kritzikratzi mentioned.

[1] https://github.com/nothings/stb

[holocronweaver]

Ran into the same problem. Perhaps this should be mentioned in the README?

I am guessing the reason the stb approach is used is so GL lib wrangling is left up to user.

[ejahandar]

what about using nvgluCreateFramebuffer (in nanovg_gl_utils.h) in multiple files, its impossible with this approach.

[KrankerApfel]

Hello,

Sorry to reopen that old issue but I struggle a lot with the same problem.

Has anything changed regarding the inclusion of nanovg_gl.h in several .cpp ? I tried to write in a first .cpp :

#include <nanovg.h> #define NANOVG_GL3_IMPLEMENTATION #include <nanovg_gl.h>

There is only one .cpp file which define that as @kritzikratzi suggests and then in my other .cpp file, I only put :

#include <nanovg.h> #include <nanovg_gl.h>

And this leads to the compile-time error C3861 : ncgCreateGL3 identifier not found from that second .cpp file.

i already tried to fix that by declaring the ncgCreateGL3 function in my file without the definition, but the linker send a LNK2019 : unresolved external symbol for the definition.

Any idea of what is happening ?

[mulle-nat]

Dude, there is no ncgCreateGL3 symbol in nanovg, so this is probably just a typo. Also maybe check #634.

[KrankerApfel]

@mulle-nat I mean nvgCreateGL3 it's not a copy-past and I make a mistake when I wrote it here, sorry.

[mulle-nat]

Try putting

#include <nanovg.h> 
#define NANOVG_GL3
#include <nanovg_gl.h>

into the second .cpp file, which doesn't define NANOVG_GL3_IMPLEMENTATION but does use nvgCreateGL3. This would solve the C3861 problem. The linker problem is some kind of goof in the setup of your project: e.g. the first .cpp that defines NANOVG_GL3_IMPLEMENTATION isn't really linked.

[KrankerApfel]

#define NANOVG_GL3

Thanks for your help! I don't really understand what happened or why adding that definition worked, but it fixed my problem.