[EMAIL PROTECTED] (Jürgen Spitzmüller) writes: > Jean-Marc Lasgouttes wrote: >> OK then. But I still think that moving the code to Text is better >> (since the code only works in text). > > So you think the lfun should be disabled if the cursor is in mathed, even if > the formula is part of a section?
No, if the lfun is only handled in Text what happens is that - Cursor::dispatch asks the top slice (mathed) whether it knows about outline ==> no - Cursor::dispatch asks the next slice (text) whether it knows about outline ==> yes! - the outline stuff is done at text level The current situation is - Cursor::dispatch asks all the slices whether they know about outline ==> no - LyXunc::dispatch tries BufferView::dispatch ==> yes, but it crashes if the cursor is in math The idea is that the lfun should be handled in the place that is the most reasonable: - LyXFunc::dispatch: all things that are not related to a particular buffer - BufferView::dispatch: all things that are global on a buffer or are able to handle transparently math or text - Text::dispatch: operations that only make sense in a text inset - InsetFoo::doDispatch: operations that only make sense in a given inset - InsetMathNest::dispatch: operations that only make sense in mathed. There should be a Buffer::dispatch for operations that work on a buffer but do not require a view, but it does not exist (actually there is one but it does not do what I would expect). JMarc
