On Thu, 2018-11-08 at 22:19 +0100, Julien Nabet wrote: > I'm not completely sure too but there's no doc about this and putting > an height or a width to a negative value (except -1 to tell it's > uninitialized) seems really weird. Well, IIRC there was something in some of the msword file formats where a negative height for something specifies that its a exact height, which a positive height specifies that its a minimum height (or the other way around) > Now I'm stuck with a third assert (see bt + gdb). > > Putting some break, I found that it was coming from > > 2707 else if ( (nType==ControlType::Listbox) && > 2708 ((nPart==ControlPart::ButtonDown) || > (nPart==ControlPart::SubEdit)) ) > 2709 { > 2710 aEditRect = NWGetComboBoxButtonRect( nPart, > rControlRegion ); > 2711 } I imagine that this is a temporary thing, the initial size of the containing panel is small/meaningless until its fully populated and then its sized to something meaningful and redrawn properly. If you're just looking for suspicious things, then there's no harm in just fixing that up locally so you can continue, but I wouldn't commit the temp hackery to master unless there is some obviously wrong thing detected. _______________________________________________ LibreOffice mailing list LibreOffice@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/libreoffice