Fixed a memory-corrupting texture bug; need help testing backends.

[sorlok_reaves]

That's not actually my cubes demo lol it's the GM Studio one that I use to show off how much faster ENIGMA renders it.
https://www.dropbox.com/s/ro00kob8723vlsb/CUBES.zip

Also please try this file.
http://pastebin.com/AeDrDsVF

That file gives me:

Code: [Select]

Graphics_Systems/OpenGL3/GL3model.cpp: In function ‘unsigned int enigma_user::d3d_model_get_stride(int)’:
Graphics_Systems/OpenGL3/GL3model.cpp:93:21: error: ‘class Mesh’ has no member named ‘GetStride’
  return meshes[id]->GetStride();

I doubt you'll be able to pack an  RGBA 4 byte color into GLfloat I already tried extensively, but be my guest :\

Ah, you're right, I was thinking of GLDouble, which I used to store colors in my polygon code... mixed 'em up.

[Goombert]

Go to that file and line and change that call to GetVertexStride() :p

Josh is also attempting to fix it right now.

[sorlok_reaves]

Go to that file and line and change that call to GetVertexStride() :p

Yeah, I'm still getting a blank screen with that.

[Goombert]

Alright, looks like I failed pretty miserably at this one, at any rate, as you can on the site tracker, Josh just committed his fix which I am testing now, please try git update and test his version of the fixes, he tested on Linux, I am going to test from Windows.

This is his commit.
https://github.com/enigma-dev/enigma-dev/commit/b539ff0bee4b085532d8d0782d2518f83022ba3a

[Josh @ Dreamland]

The fix is mostly duct-tape. Those templates need moved to their own header, as they are general purpose (which is why I wrote the signed-int set as well). The color_t type also needs used in more places. The point for now was to make sure it was the same size as the union. Switching gs_scalar to double might yet break it; at very least, it will waste four bytes.

[sorlok_reaves]

The fix is mostly duct-tape. Those templates need moved to their own header, as they are general purpose (which is why I wrote the signed-int set as well). The color_t type also needs used in more places. The point for now was to make sure it was the same size as the union. Switching gs_scalar to double might yet break it; at very least, it will waste four bytes.

Works for me. Gotta say, I'm impressed by the way the union works. Anyway, GL3 is working great now. Thanks for the fix.