Alfredo Braunstein wrote: > All insets, not only top-level (but you probably meant that).
Ahh yes, you're right. However, note that the code *use to* iterate over all insets with the proviso that the entire contents of a math inset were treated as a single whole. That is, all those bit-of-math insets weren't really insets at all. The real inset was an InsetFormula that could contan only bit-of-math insets. I have no idea if that logic is still correct. Are the bit-of-math insets proper insets now? -- Angus
