variables that stored the coordinates of the cursor prior to the
current mouse move. They were un-initialised only for the first mouse
move after a mouse button down event.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5651 c8812cc2-4d05-0410-92ff-de0c093fc19c
There still seems to be a slight issue when showing the Project Properties for the first time on a blank project (i.e. it scrolls *near* the WGS 84 entry, but not quite on top of it), but I'll leave that for another time. Other situations now seem to work OK.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5642 c8812cc2-4d05-0410-92ff-de0c093fc19c
include the offending sql in the error dialog box.
Also fix the remaining part of ticket #131 - the code wasn't coping
correctly with blank data fields.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5637 c8812cc2-4d05-0410-92ff-de0c093fc19c
to the spatial reference system database (it was never returning a
result because the proj4 string didn't have '+no_defs' on the end)
git-svn-id: http://svn.osgeo.org/qgis/trunk@5635 c8812cc2-4d05-0410-92ff-de0c093fc19c
When a new WMS layer is opened and it is the first layer on the canvas, the project projection is now copied from that used when selecting that WMS layer.
Previously this was hard coded to EPSG:4326 / WGS 84.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5627 c8812cc2-4d05-0410-92ff-de0c093fc19c
This is because QgsProjectionSelector::setSelectedSRSID() now works as advertised when called externally.
It didn't work before, because setSelectedSRSID() did not make sure the tree widget was pre-populated with the projection list.
This commit is a pre-requisite to fixing #11. This is so that we can actually perform the experiment to prove future bug fixes against #11.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5618 c8812cc2-4d05-0410-92ff-de0c093fc19c
.shp file (if it is read-only, qgis then tries to open it read-only,
so the OGR error message is unnecessary).
git-svn-id: http://svn.osgeo.org/qgis/trunk@5617 c8812cc2-4d05-0410-92ff-de0c093fc19c
instead of separate icons in separate columns. Still only applies to
the actual layer file entry in the legend, but removes the need for
the extra columns. However, I'm hopeless at icon drawing...
These changes are a variation on the suggestions in ticket #191.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5616 c8812cc2-4d05-0410-92ff-de0c093fc19c
I'm not sure why this functionality hasn't been done before, as it seemed so trivial to implement?
The main benefit to its implementation is that in large layers, the user doesn't have to wait for the whole layer to draw before he can restart work on a project (which is particularly important since we currently can no longer hit Esc to halt a redraw).
git-svn-id: http://svn.osgeo.org/qgis/trunk@5613 c8812cc2-4d05-0410-92ff-de0c093fc19c
This appears to be because of the use of QWidget::removeChild in QgsComposer::removeWidgetChildren, which is a Qt3 idiom (it doesn't appear in the Qt4.1 API). A QWidget::setParent(0) idiom is now used instead.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5611 c8812cc2-4d05-0410-92ff-de0c093fc19c
bug whereby the first byte after the wkb array was being set to \0
git-svn-id: http://svn.osgeo.org/qgis/trunk@5609 c8812cc2-4d05-0410-92ff-de0c093fc19c
With the Identify tool on a WMS layer, now send a query to the WMS server *only* on those layers that the WMS server had already declared queryable. This avoids the server returning an error on un-queryable layers.
Also, QgsWmsProvider::identifyAsHtml() is renamed to QgsWmsProvider::identifyAsText() to better identify what it does. Commentary, and other dependent variable and function names have also been updated to suit.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5606 c8812cc2-4d05-0410-92ff-de0c093fc19c
Unfortunately I don't have access to a proxy to test, so if I could get some independent verification it would be appreciated.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5605 c8812cc2-4d05-0410-92ff-de0c093fc19c
Now, the fact the WMS raster provider was used, the layer names, layer styles and image formatting is stored in the project file (and restores OK). However upon initial testing the extent doesn't appear to be restored. I will need to address that later.
git-svn-id: http://svn.osgeo.org/qgis/trunk@5603 c8812cc2-4d05-0410-92ff-de0c093fc19c
