Andre Poenitz wrote: >> We should do the same for ParIterator. > > You could do me a favour by having a look at ParIterator and merge this > with DocumentIterator. Of course a slim wrapper around DocumentIterator > would be fine.
It is less trivial than I thought: DocIterator is inset-based and ParIterator is ParagraphList-based. So we have problems for instance in CutAndPaste, where the text is stored as a ParagraphList and not an inset. Probably a solution would be to iterate over (for textclass changes, back pointers) only after inserting the text into the document. Alternatively, we could put the ParagraphList into a temp inset. What would be better I don't know. Alfredo
