I must confess that, so far, I have been working on getting the original QtMmlWidget's MathML renderer to work within Qwt (which is now the case). From there, I have been working on fixing some issues I have found with the original MathML renderer (still working on it).

This being said, I agree that a floating point based approach would be better. Maybe I should go through the code and, who knows, it might mean I don't have 'patch' certain things. Then again, there are some factors that were just plain wrong, so I had no choice but to 'fix' them. For example, say that you have x with i as a sub-script and j as a super-script. In this case, the MathML render uses g_script_size_multiplier to scale those scripts. However, the original value of g_script_size_multiplier is such that i and j would always overlap, hence I modified that value to 0.5 (from 0.7071) and as expected now everything is fine.

Anyway, it's still very much work in progress, but I am hoping to be done with it within the next few days. (I can't afford to spend too much time on this.) From there, I leave it to you to incorporate 'my' version of the MathML render to Qwt or not. What is certain is that you are better placed than I am to test the renderer for use within Qwt (my use of Qwt is somewhat limited).