Recent change in MapView creating problems in projections

[pieren]

A recent change in Mapview.java (rev.2450) by "jttt" introduces a
problem with the Lambert projections.

The change is the following in Mapview @Override public void paint(Graphics g) :
(before)

Graphics2D tempG = offscreenBuffer.createGraphics();
tempG.setColor(Main.pref.getColor("background", Color.BLACK));
tempG.fillRect(0, 0, getWidth(), getHeight());
for (Layer l: getVisibleLayersInZOrder()) {

l.paint(tempG, this);

}
for (MapViewPaintable mvp : temporaryLayers) {

mvp.paint(tempG, this);

}

(and after)

Graphics2D tempG = offscreenBuffer.createGraphics();
tempG.setColor(Main.pref.getColor("background", Color.BLACK));
tempG.fillRect(0, 0, getWidth(), getHeight());
Bounds box = getLatLonBounds(g.getClipBounds());
for (Layer l: getVisibleLayersInZOrder()) {

l.paint(tempG, this, box);

}
for (MapViewPaintable mvp : temporaryLayers) {

mvp.paint(tempG, this, box);

}

The problem is this new call of getLatLonBounds() with random east,
north values the first time a new layer is added. It's not exactly random as the values are based on the screen width and height !

The solution is not to ignore those values in the projection implementation when they are outside the box otherwise I cannot warn the user when he is really working outside the projection limits.

This problem is also independent of the subprojection things to do. It is currently blocking the cadastre-fr plugin working only with the concerned projections.

[jttt]

The value is not random (and it's not a bug afaik), it's bbox arount default center, calculated by NavigatableComponent.calculateDefaultCenter(). Default center is calculated as:

        Bounds b = Main.proj.getWorldBoundsLatLon();
        double lat = (b.getMax().lat() + b.getMin().lat())/2;
        double lon = (b.getMax().lon() + b.getMin().lon())/2;

        return Main.proj.latlon2eastNorth(new LatLon(lat, lon));

Problem here is that first call of MapView.paint is before any data are loaded. As a workaround I can check whether there is any non empty OsmDataLayer and if not then use whole world bounds instead of value calculated by getLatLonBounds().

But that seems a bit hackish, I would preferer if all Projections returned polygon where the projection is valid and center point of that polygon.

[stoecker]

When the world bounds are fixed to be proper for all the relevant projections, then this code should return valid coordinates.

[jttt]

What does the value returned by getWorldBoundsLatLon() means? Is it bbox where projection is valid?

[stoecker]

Yes.

[pieren]

In ticket #3989, I added a better support of getWorldBoundsLatLon() in the patches to apply.

[pieren]

I've attached a patch file "NTV2Grid.patch". Sorry about that.

[pieren]

Replying to pieren:

I've attached a patch file "NTV2Grid.patch". Sorry about that.

Ooops, wrong ticket.

[jttt]

Can this ticket be closed?

[pieren]

Should be fixed with #3989