Juergen Spitzmueller wrote: >> Does the following one-liner patch help? I think this is a consequence >> of the fix of bug 1415, which increased the value of num_units. > > Thanks, Jean-Marc. I'll try it when I am back home (Thursday).
Actually, I just tried it and it fixes the crash. As in 1.3.x, invalid data is now just omitted (replaced by 0 [default unit]). That's ok for the core. I try to implement some validators for qt (checkedwidgets), so that the user cannot apply invalid data at all (and gets a visual clue). xforms already has this. I have no cvs access ATM. Can you please commit your fix and mark bug 1682 fixed? Many thanks, Jürgen BTW: In 1.4, we now tranlate local values ("1,2") correctly (to "1.2"). This happens in widgetsToLength, but unfortunately only for non-glueLengths (glue Lengths with localized values are not validated). We could do something like the following in widgetsToLength: if (isValidGlueLength(fromqstr(length))) { string len = subst(fromqstr(length), ',', '.'); return len; } but I have a feeling that this is not the ideal solution. Do you have an idea on how to do the translation in LyX(Glue)Length?