On Tuesday 18 January 2011, Cyrille Berger Skott wrote:
> Hi,
>
> I had a similar idea. But instead of testing the exact rendering, which is
> rather fragile, my idea was to implement a special QPaintEngine (and
> associated QPaintDevice), that would record as text the drawing commands.
Doesn't QPicture do something like that already? Or else we might compare the
svg.
>
> The idea is that the actual drawing on screen is not up to our
> responsability,
> it would be a Qt matter, what is important from a calligra point of view is
> that the call to QPainter's API is correct.
>
> The way I see it would be outputing text that would look like this:
> "LINE 12.0 -12.1 34.3 2.3 QPEN(255,0,0)"
>
> Which means that a red line has been drawn from (12.0,-12.1) to (34.3, 2.3).
>
> And in the QPaintEngine it would be implemented as:
>
> void QPaintEngine::drawLines ( const QLineF * lines, int lineCount )
> {
> for(int i = 0; i < lineCount; ++i)
> {
> addText( QString("LINE %1 %2 %3 %4 %5")
> .arg(lines[i]->x1())
> .arg(lines[i]->y1())
> .arg(lines[i]->x2())
> .arg(lines[i]->y2())
> .arg(currentPenText()) );
> }
> }
>
> And then you could use this QPaintEngine/QPaintDevice with QPainter, and
> basically test anything that need to be drawn. That could range from unit
> testing KoShape::paint, to testing the rendering of document.
>
> This approach also has the advantage that it should be platform independent,
> meaning that anyone could run the test suite, and it would not break on
> update
> of the testing platform.
>
> (on a side note, my intention was to have Hanna works on this when she is
> more
> at ease with Calligra code, but if you want to take that idea, I would not
> mind, and will find her something else to do :) )
>
>
--
Boudewijn Rempt | http://www.valdyas.org, http://www.krita.org
_______________________________________________
calligra-devel mailing list
calligra-devel@kde.org
https://mail.kde.org/mailman/listinfo/calligra-devel
