[kde-artists] Re: Oxygen Tab Widgets.

On Tuesday 19 April 2011 16:06:30 Alexis Ménard wrote:
> Hello Hugo,
> 
> So I looked a bit inside Oxygen.
> 
> I did a patch (which is wrong because it does not fix all the cases
> for QTabWidget) but seems to give almost the same look as before but
> fix the scrolling issue by avoiding overlap.
> 
> Attached 3 screenshots:
> 
> - The original one (today) (number1)
> - With your line commented (number2)
> - With my patch (number3)
> 
> Also if I comment the line you said nothing happen because the overlap
> is used somewhere else. If I put no overlap like in the second diff
> attached a margin appears on top of the content.
> 
> What do you think?
> 
Very helpful !
In fact in the meanwhile, I built a debug version of qt and could reproduce 
the issue. I started to look into it, and your patches, investigations will 
certainly help. I should be able to build on your third patch.

(PS: oxygen-demo is a tiny app that allows to debug these kind of things. The 
tab called "tabwidgets" notably, allows to test all possible cases for tab 
widgets (with/without frame, all 4 orientations, with/without corner buttons, 
and in reverse layout mode ... many combinations. Hence the pain)

Hugo

> On Mon, Apr 18, 2011 at 7:05 AM, Hugo Pereira Da Costa
> 
> <[email protected]> wrote:
> > On Monday 18 April 2011 12:04:21 Hugo Pereira Da Costa wrote:
> >> On Monday 18 April 2011 11:49:52 Hugo Pereira Da Costa wrote:
> >> > Hello Alexis,
> >> > 
> >> > So I did some digging, and well, I'm not sure I have the skills to fix
> >> > myself either.
> >> > 
> >> > This "overlap" thing, corresponds to what QStyle::pixelMetrics(
> >> > PM_TabBarBaseOverlap ) should return (so Qt knows about it).
> >> > 
> >> > it is used in these two methods
> >> > 
> >> > Oxygen::Style::tabWidgetTabPaneRect
> >> > Oxygen::Style::tabWidgetTabContentsRect
> >> > 
> >> > which are what's called (by Qt) via
> >> > 
> >> > QStyle::subElementRect( SE_TabWidgetTabContents )
> >> > QStyle::subElementRect( SE_TabWidgetTabPane )
> >> > 
> >> > Looking at QCommonStyle implementation for the two above, well, they
> >> > don't differ 'much' in terms of metrics handling. (in fact I remember
> >> > copying them, and "tuning" them with +/-1 pixels here and there to fix
> >> > some bugs when dealing with corner cases of QTabWidget/KTabWidget
> >> > handling).
> >> > 
> >> > QTabWidget is (by far) the most painful widget to deal with in styling
> >> > land and I've already spend hours (tenth's of hours) on it. (notably,
> >> > there is no existing Qt Style that does not have one bug or another
> >> > with it, in one possible or the other).
> >> > 
> >> > I'm ok to revisit it in view of the problem you raise, but this will
> >> > take time (and my hand are pretty full at the moment).
> >> > 
> >> > Feel free to file an actual bug report to oxygen (style) with as much
> >> > detail as you can so that it does not get lost in translation.
> >> > 
> >> > Cheers,
> >> > 
> >> > Hugo
> >> 
> >> PS: I don't have (yet) a Debug build of Qt.
> >> Could you check if the bug persist if you comment the two lines:
> >> 
> >>             case SE_TabWidgetTabContents: return
> >> tabWidgetTabContentsRect( option, widget );
> >>             case SE_TabWidgetTabPane: return tabWidgetTabPaneRect(
> >> option, widget );
> >> 
> >> in oxygenstyle.cpp ?
> >> this should fallback to QCommonStyle implementation, will raise more
> >> bugs in some applications, but might give me more clue about what's
> >> going wrong.
> >> 
> >> (note: maybe you can also try just commenting the first)
> > 
> > argh. Sorry. I meant: just the second, namely
> > case SE_TabWidgetTabPane: return tabWidgetTabPaneRect( option, widget );
> > 
> > They should be around line 880 and 881
______________________________________________________________________________
[email protected] |  https://mail.kde.org/mailman/listinfo/kde-artists