quick fix, crash after too many calls

[Lectem]

I added a pointer check for each sf2d_draw_* function so that app doesnt crash but stop drawing instead.

I don't understand why you would use a pool if it's not to draw it all at once though...

[xerpi]

Good one, basically I use a temporary memory pool to allocate vertices because maybe you only want to draw a rectangle on one frame, and not on the next, so I use this fast pool (mallocing on that pool is just moving an offset) to be able to do that without overhead, and then I set the offset to 0 at the start of each frame.
Btw I think:

if (!vertices) return;

Would be better (if you don't mind).

[Lectem]

I still think that it is more or less useless. I mean, having a simple pool of union of sf2d_vertex_pos_tex and sf2d_vertex_pos_col would suffice for this, the only problem would be if GPU_DrawArray(GPU_TRIANGLE_STRIP, 4); was asynch and that we changed values too fast

[Lectem]

If you really need a pool, make it a cyclic one, if you reach the end, reset it so that we can still draw stuff

[xerpi]

The problem is that the GPU_DrawArray simply adds a command to the GPU command queue, so we must not delete the vertices until we call:

    GPU_FinishDrawing();
    GPUCMD_Finalize();
    GPUCMD_FlushAndRun(NULL);

[Lectem]

I wonder if we couldn't simply do

GPUCMD_FlushAndRun(NULL);
gspWaitForP3D();
GPUCMD_SetBufferOffset(0);

or

GPUCMD_Run(NULL);
gspWaitForP3D();
GPUCMD_SetBufferOffset(0);

so that the the cmd buffer would be treated and just continue issuing new commands.
Anyway, I think you could close the pull request as this subject is another issue

[xerpi]

Right, closed.