On Wed, Apr 06, 2005 at 03:30:25PM +0200, Helge Hafting wrote: > The general problem: where should the selected stuff go, if the user > applies something that have more than one input box? > For a fraction, the stuff goes on top which is ok. For a > n-root it goes into the n-part, which is less useful.
'Auto mobing cell contents' always moves the selected stuff into the first cell of a MathNestInset, which is the '3' part in a cube root. > People usually have lots of text under the root sign and a single > numeral (or letter) as "n", so "under the sign" is the most useful > place to autopaste. For the cube root special case the user don't > think of the "3" as editable at all. I did not know we had a cube root special case anywhere at all. Is this the Qt-frontend? If so, I'd rather (a) drop that special case or (b) provide a special cube-root-inset, but not modify the general MathInset code in any way. When I think about it, I prefer (a). I've never used cube roots to a degree rectifying any special handling over an nth root. > One can also mark 3+4 inside 1+2+3+4+5 and press superscript or > subscript. The resulting 3+4 raised to some power is a bit confusing, > mathematically it just looks like 3 alone plus 4 raised to some power, > but mathed-wise it is (3+4)^ without the parentheses. This is more or less intentional even if it has the possibility of being misused. But then, this 'feature' is available in TeX and I do not want to have any kind of automatic operator precedence handling in mathed (except in the CAS code) the will add fragility and failures in corner cases. If you create a superscripted 3+4 you get one because you asked for it. It's as simple as that. Andre'
