[rlgl] Allow tint colors in GL_LINE (wires) and GL_POINT (points) draw modes on OpenGL 1.1

[meisei4]

Decided to make a PR because the fix was a little more involved than i felt i could show on discord, and i wanted to show more cleanly where in the code im observing the fix behavior.

Issue

In the Opengl 1.1 context the colors array is getting enabled incorrectly during wire and point mode i think

Desired behavior: DrawModelWires(..., Color tint), should draw the wire model with the wires being colored by "tint"

Observed Behavior: loading an obj file and setting it to draw the model fill with WHITE and wire frame any color other than WHITE fails.

Workaround: if you nullify the .colors field in the mesh before drawing with wires or points, you can set the color in the draw function

this PR's solution: during DrawMesh, find the current GL_POLYGON_MODE and from there do not enable the GL_COLOR_ARRAY if you are in modes of GL_LINE or GL_POINT, this allows for the draw functions tint to apply without the workaround.

NOTE: I also added rlSetPointSize(float size) (and getter) to allow for glPointSize(size), because my eyes could not debug point mode and color well without this feature of increasing point size (its also a feature i would like to have for debugging in opengl1.1)

Concerns:

I am wondering if this code change has any consequences outside of what i believe to be normal behavior matching the other opengl modes now.

Images of current behavior (will only draw white wires, even when undesired -- e.g. on top of an also white model fill)

Images of desired behavior (custom point and wire colors over the main model color)

[raysan5]

I'm afraid this is an undesired approach...

[raysan5]

This is not following raylib coding conventions.

[meisei4]

updated to have the curly brace conform and added comments the same as Line width (I am basing the change mostly on line width)

[raysan5]

@meisei4 I see the issue but the proposed approach should be reviewed. rmodels never calls gl*() functions directly

[meisei4]

updated with a simple null check, even with the wire and points poor solution it didnt solve the fill mode, actual solution is just if(mesh.colors), just like the rest of the mesh fields

[meisei4]

final result with the fix:
BLUE fill, RED wires, GREEN points

and then with a material example

[meisei4]

Updating this because actually i falsely assumed i had rebuilt everything correctly. The issue still remains with OpenGL 1.1 even with the mesh.colors null check.

The reason in my case is because LoadOBJ will always allocate WHITE/255 into the colors field for the mesh:

raylib/src/rmodels.c

Line 4422 in 84e2cbc

model.meshes[i].colors = (unsigned char *)MemAlloc(sizeof(unsigned char)*vertexCount*4);

raylib/src/rmodels.c

Line 4490 in 84e2cbc

for (int i = 0; i < 4; i++) model.meshes[meshIndex].colors[localMeshVertexCount*4 + i] = 255;

The workaround is to just set the mesh.colors to NULL after you load the obj

apologies for the sloppiness, I am still learning, and i imagine there is a more appropriate way than just inserting the opengl context macro into the LoadOBJ (Also I have a feeling this might be an issue elsewhere where for opengl 1.1 the colors can be a bit tricky to deal with smoothly in comparison to other opengl contexts)🙇‍♂️.

[raysan5]

I'm afraid glPointSize() is not available on OpenGL ES 2.0. Please, could we avoid adding this function?

[meisei4]

I was under the impression that it would be ok, because of the equivalent pattern for WireMode (and polygon mode in general according to comments i believe) being gated by context macros:

https://github.com/raysan5/raylib/blob/master/src/rlgl.h#L1968

Would it be ok to keep it if we add the appropriate version context checks like with WireMode?🤔

[raysan5]

yeah, we can use same approach

[raysan5]

I'm afraid GL_POINT_SIZE is not available on OpenGL ES 2.0. Please, could we avoid adding this function?