Andre Poenitz wrote:
>> I think this has been reported before...
>
> Indeed. And I haven't had time to fix it so far.
>
> The problem is most likely an off-by-one (or two) bug in accessing
> the cells of the script inset. A while ago, cell 0 was always the
> subscript, cell 1 the superscript and cell 2 the base and there was
> some extra code to handle 'missing' super/subscripts. Such code was
> also necessary in the DocIterator, and it clearly doesn't belong
> there.
>
> So now we have the base (which must exist) in cell 0. If there are
> both sub- and superscript, the superscipt is in 1 and the subscript
> in 2, and if there is just one of them, it's in cell 1. 'nargs'
> returns now 2 or 3 (not always 3 as formerly)
>
> The access should go through some wrapper functions there
> (up/down(?), don't have sources here...) and it looks like I f*
> something up there.
Here's one error, although I think that nargs() == 1 is possible which
would cause the 'c' part of 'a?b:c' to fail also...
MathArray const & MathScriptInset::down() const
{
- return nargs() == 2 ? cell(2) : cell(1);
+ return nargs() == 3 ? cell(2) : cell(1);
}
MathArray & MathScriptInset::down()
{
- return nargs() == 2 ? cell(2) : cell(1);
+ return nargs() == 3 ? cell(2) : cell(1);
}
Here, should you test nargs()?
void MathScriptInset::notifyCursorLeaves(idx_type idx)
{
MathNestInset::notifyCursorLeaves(idx);
// remove empty scripts if possible
- if (idx == 2 && cell(2).empty()) {
+ if (idx == 2 && nargs() > 2 && cell(2).empty()) {
removeScript(false); // must be a subscript...
- } else if (idx == 1 && cell(1).empty()) {
+ } else if (idx == 1 && nargs() > 1 && cell(1).empty()) {
if (nargs() == 2) {
cell_1_is_up_ = false;
cell(1) = cell(2);
cells_.pop_back();
} else if (nargs() == 1) {
cells_.pop_back();
}
}
}
--
Angus
