On Sat, Mar 22, 2003 at 01:08:04PM +0100, Alfredo Braunstein wrote: > patch2 to lyxrow.C. On lastPos we returned 0 if the row's paragraph is void. > But zero is the right value to return if the paragraph has 1 char, so this > is pretty wrong. Moreover, returning 0 fires a bogus getChar request on > computeBidiTables (activate rtl, file->new, crash). I've put it to return > -1 without problems so far.
No !! Wrong wrong wrong. It's meaningless to return -1. Simple fact is: routines must be able to deal with empty paragraphs. Any spots that can't are simply bugs. And yes, I specifically changed this to return 0 not -1, because *it was causing bugs*. computeBidiTables must be fixed. regards john