Well, I didn’t think you had been harsh and I also not intended to be it. So let’s forget that.
I never saw any guy that would write a translation matrix different than this:
1, 0, 0, tx
0, 1, 0, ty
0, 0, 1, tz
0, 0, 0, 1
I’m creative so I could think of that:
0, 1, 0, 0
tx, 0, 1, 1
1, tz, 0, 0
0, 0, 0, ty
I’m quite laughin a bit now…
But, to get to what I mean. There is kind of STANDARD REPRESENTATAION OF A MATRIX, AND STANDARD OPERATIONS APPLYING TO THAT REPRESENTATION. If not, any book would contain different ways of drawing matrices. I bet you will never find any other representation of that column based (the first one of cause) translation matrix.
Now if you lay the opengl pattern over it:
0 4 8 12
1 5 9 13
2 6 10 14
3 7 11 15
You can find the memory offsets of the elements when you want to put your drawn matrix into memory. I wouldn’t understand if anybody drifted off from that easy line.
And in case, I don’t mean to offend you. It’s just that my theoretical question go in the background with this discussion… And I’m an ignorant!