> i'm thinking that the backspace operation should have deleted both
> the section and the block rather than just the second section. that
> is after the backspace the dump should (give or take a pointer value)
> look like the first dump.
after i sent this i remembered that i had put in code to silently
convert a 'delete first block in section' to a 'delete the section
and let blocks be assimilated by previous section'. so please disregard
this suggestion. cursor motion and editing operations around section
breaks are probably a little ambiguous/arbitrary until we get "show-invisibles"
working....
> i'm going to hold off on tracking this down myself ....
i tried, really i tried, i tried not to just go find it and kill it -- i
managed to wait all of an hour or so before the unseen forces pulled me
back to the keyboard....
in fl_DocSectionLayout::doclistener_deleteStrux() there's a loop which moves
blocks from one section to another (calling removeBlock() and addBlock()).
it looks like fl_SectionLayout::addBlock() should set the pointer in the
block (looks like we need another method on fl_BlockLayout to do this for us)
and (to be safe) removeBlock() should probably clear it. (i haven't tested
this, but it appears that fl_SectionLayout::insertBlock() method does cause
it to be set via the fl_BlockLayout constructor, so i think that's right.)
jeff