Am Freitag, den 10.03.2017, 15:17 +0100 schrieb Jean-Marc Lasgouttes: > Le 10/03/2017 à 14:59, Jürgen Spitzmüller a écrit : > > > A way to fix the coverity issue would be to assert on p>=0, but I > > > have > > > to ask: how do we know that p!=-1? > > > > AFAICS it can actually be -1, if the filter string is not found. So > > we > > probably should not assert here, but leave the loop. > > > > So > > > > - LASSERT(p != -1, /**/); > > + if (p < 0) > > + break; > > But this would mean that I can make it assert easily, right? I did > not > manage to do that.
Right, actually it seems to be impossible. > > My experience is: > - qt4: no crash, no underline > - qt5: no crash, correct underlines. > > Are both situations expected? No, it should also underline with qt4 (I do not have a qt4 build here to test myself, but I remember it used to work). > > > The condition above, AFAICS, looks for consecutive matching strings > > which are then concatenated, i.e. > > > > <u>found</u>found... -> <u>foundfound</u>... > > > > If the condition does not match, the matching string only applies > > to > > the actual hit, so it is wrapped in <u>...</u> > > That is what I figured out. I seems to me that it could be done in > an > easier way, although I do not know which one yet :) Me neither. > > > An easy question to en the message: the lone "cc_->filter();" > > > statement > > > before the loop does nothing AFAIU. Was it supposed to be > > > something > > > else? > > > > I suppose it can be removed (also in LayoutBox.cpp). > > I missed that one. Is there a reason why these two classes are > separate? Laziness. They should eventually be merged. IIRC LayoutBox differs in some respects (or is tied to layouts), but I think it could inherit and modify CategorizedCombo. Jürgen > > JMarc >
signature.asc
Description: This is a digitally signed message part