Currently working on a patch to address deprecations that trigger -Werror with Qt 5.14.1

[Scott Kostyshak]

On Fri, Mar 20, 2020 at 10:17:51AM +0100, Jean-Marc Lasgouttes wrote:
> Le 20/03/2020 à 03:41, Scott Kostyshak a écrit :
> > > The entry in the change log hints (from my understanding) that it is a
> > > renaming, but the blog article I linked to suggests things could be more
> > > complicated.
> > 
> > Looking at the code, it does seem to be equivalent:
> > 
> >    qreal QFontMetricsF::width(const QString &text) const
> >    {
> >        return horizontalAdvance(text);
> >    }
> 
> Good.
> 
> > So it seems it would not be a regression for us to rename every instance
> > to horizontalAdvance(), although I would have to check the equivalence
> > for all variants of the method. However, I guess the idea here is it's a
> > good opportunity to catch a mistake in our code?
> 
> Yes, we should catch the cases where we need the total width(). I have
> issues to  find time though, due to our confinement status in France. 5
> people in a flat all day is a bit complicated :)

Ouch, I'm sorry, JMarc. That must be very uncomfortable and stressful.

> > Let's take the first case then, which is CCItemDelegate::drawCategoryHeader:
> > 
> >    // draw the centered text
> >    QFontMetrics fm(font);
> >    int w = fm.width(category);
> >    int x = opt.rect.x() + (opt.rect.width() - w) / 2;
> >    int y = opt.rect.y() + 3 * fm.ascent() / 2;
> >    int left = x;
> >    int right = x + w;
> >    painter->drawText(x, y, category);
> > 
> >  From what I understand, this code wants to center the category header.
> > Although I've read up on the differences of boundingRect().width() and
> > horizontalAdvance(), I don't know which one is correct here. In some
> > sense, we are not really drawing text after the string---we just want to
> > know where to start drawing it so that it is centered. So that makes me
> > think boundingRect().width() would be more appropriate. On the other
> > hand, I wonder if horizontalAdvance() would provide a centering that is
> > more natural to the eye. If I had to guess, I would say that
> > boundingRect().width() is the correct choice here: I'm picturing a
> > string with weird bearings on the first and last character, and I think
> > my eyes would prefer for the bearings to not be taken into account while
> > centering. Is my approach to thinking about this reasonable?
> 
> I would say that we don't care, so width() is better since is is shorter and
> easier to understand :)

Just to double-check: By width() you mean boundingRect().width(), and
you are saying it is shorter than horizontalAdvance() because we do not
have to condition on the Qt version, right? I agree that
boundingRect().width() is easier to understand than horizontalAdvance().

Scott
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: not available
URL: <http://lists.lyx.org/pipermail/lyx-devel/attachments/20200320/de93aafc/attachment.asc>