AW: [Qgis-developer] "Fix problems with incorectly filled polygons"issues

[Hugentobler Marco]

I think the thing that brings your performance down is transparency. Editing markers and measuring rubber bands both use transparency (and GRASS edits don't). 
FPWIPF uses QPixmap to render (and is rendered on the X-server). If FPWIPF is disabled, QImage is used which afaik is pure software rendering.


Regards,
Marco





-----Ursprüngliche Nachricht-----
Von: qgis-developer-bounces at lists.osgeo.org im Auftrag von Maciej Sieczka
Gesendet: So 13.04.2008 20:08
An: qgis-developer at lists.osgeo.org
Cc: qgis-user at lists.osgeo.org
Betreff: [Qgis-developer] "Fix problems with incorectly filled polygons"issues
 
Hi,

I have a couple of questions and comments regarding "Fix problems with
incorectly filled polygons" (let's shorten it to FPWIFP to save key
strokes).

1. If FPWIFP is set, rendering is very slow *during digitizing*. Same
shapefile or PostGIS layer with FPWIFP set on, renders in 20 seconds,
whereas with FPWIFP unset about 0.3 s - over 66 times (!) less.

However, after quiting the digitizer, it is opposite. Same vector layer
renders 2 times longer, when FPWIFP is unset.

Are these features or defects?

2. I use QGIS through NX a lot. If I want to use measure area or length
tools, or to digitize PostGIS/OGR vectors, I have to set FPWIFP on, as
otherwise the geometries I draw on the screen are not visible or
garbled hwile drawing them. The attached FPWIFP_off_digitise1.png shows
how garbled is a polygon during digitising, and FPWIFP_off_digitise2.png
shows the same polygon when finished.

Ineterestingly, in GRASS Edit such artifacts never crop out, no matter
if FPWIFP is set or not. Why the difference? Could PostGIS/OGR editor
use the same method for rendering as GRASS Edit, so that display
artifacts in NX sessions would be avoided?

3. Using QGIS through NX I often experience sudden session shutdowns, if
FPWIFP is set, with specific vector layers, that have many polygons with
numerous vertices (eg. like raster to vector output). Same layers
display just fine when FPWIFP is off. More details in my other post [1].
I know this is not a QGIS issue but rather X or NX one, but I'm
wondering if described NX shutdowns could be prevented by changing
something in QGIS rendering code?

The analysis on [2] suggests that the problem might crop out when QGIS
tries to render polygons which are smaller than one pixel at the current
scale. Maybe QGIS could skip rendering them at all? Would skipping
features invisible at the given scale improve the rendering performance
in general too?

[1]http://www.nabble.com/forum/ViewPost.jtp?post=16433182&framed=y
[2]https://bugs.freedesktop.org/show_bug.cgi?id=15307#c2

Maciek