allow for non-discrete resolution/zoom levels (fractional zoom)

This needs solid testing, and I welcome comments on the approach - for an example, see http://dev.openlayers.org/sandbox/tschaub/fractional/examples/fractional-zoom.html

Tim, I tested this with WMS layers and it works like a charm. Can't test it with commercial layers though.

One question though: it appeared like this won't work if the map has a scales array, so I had to use minScale and maxScale to get it to work. But I guess that's to be expected.

Tim, I love this change and I think it will be an important addition for several projects using OpenLayers. Mike will be testing this with Fusion, this week hopefully, and providing feedback as required.

Thanks Paul. Yeah, it needs some real field testing. Will be good to get your feedback.

I think this is a fairly non-invasive change to get this functionality. The things it does change aren't liable to haunt us (in my opinion). If an app designer sets map.fractionalZoom to true, they should know not to access layer.resolutions[zoom] directly.

I added a map.allowFracZoom method. I don't necessarily like this, but it makes things degrade well for commercial layers. The part I don't like is the check for a RESOLUTIONS constant. But I like less the alternative of modifying anything in the FixedZoomLevels layers.

(As an aside, I'm also curious about the this.resolutions != null checks in FixedZoomLevels. Since all the classes in our lib that inherit from FZL have a RESOLUTIONS constant, and initResolutions sets resolutions, I don't immediately see when this condition would ever be met. But I'll leave that for elsewhere.)

Replying to bartvde:

Tim, I tested this with WMS layers and it works like a charm. Can't test it with commercial layers though. One question though: it appeared like this won't work if the map has a scales array, so I had to use minScale and maxScale to get it to work. But I guess that's to be expected.

Hmm. Maybe I need more info. If I create a map like

            var options = {
                scales: [442943842.5, 221471921.25, 110735960.625, 55367980.3125, 27683990.15625, 13841995.078125, 6920997.5390625, 3460498.76953125, 1730249.384765625, 865124.6923828125, 432562.34619140625, 216281.17309570312, 108140.58654785156, 54070.29327392578, 27035.14663696289, 13517.573318481445]
            };
            map = new OpenLayers.Map('map', options);
            var wms = new OpenLayers.Layer.WMS(
                "OpenLayers WMS",
                "http://labs.metacarta.com/wms/vmap0",
                {layers: 'basic'}
            );
            map.addLayer(wms);

everything works as I'd expect. How are you setting your scales array?

Replying to tschaub:

I think this is a fairly non-invasive change to get this functionality. The things it does change aren't liable to haunt us (in my opinion). If an app designer sets map.fractionalZoom to true, they should know not to access layer.resolutions[zoom] directly. I added a map.allowFracZoom method. I don't necessarily like this, but it makes things degrade well for commercial layers. The part I don't like is the check for a RESOLUTIONS constant. But I like less the alternative of modifying anything in the FixedZoomLevels layers. (As an aside, I'm also curious about the this.resolutions != null checks in FixedZoomLevels. Since all the classes in our lib that inherit from FZL have a RESOLUTIONS constant, and initResolutions sets resolutions, I don't immediately see when this condition would ever be met. But I'll leave that for elsewhere.)

---

Tim,

Some questions/comments on your patch:

• the getResolutionForZoom method of the Map class checks whether that a method of the same name is defined on the base layer. Since the getResolutionForZoom method is part of the generic Layer class, I don't see how this condition wouldn't be met.

• I'm also not a big fan of the RESOLUTIONS check in allowFracZoom. Why not define the new property allowFracZoom in the Layer class and have the FixedZoomLevels class set this property to false? In that way, layers that don't permit fractional zoom can advertize it - the TileCache layer might be of those classes by the way, and this isn't caught by your RESOLUTIONS check (I'm not too familiar with the TileCache layer and be cautious with my assessments :-)

• Line 820 of the patch, you use the i variable while it was defined outside the current scope in the for loop above.

Hi Tim, another question about this. Why is map.zoomTo not adapted to also take in floats? When I want to zoom now to a fractional level I have to do the following detour:

var zoomLevel = this.map.zoom + (deltaY/this.zoomStopHeight);
var scale = OpenLayers.Util.getScaleFromResolution(
    this.map.getResolutionForZoom(zoomLevel), this.map.units );
this.map.zoomToScale(scale);

Whereas I would want to do:

var zoomLevel = this.map.zoom + (deltaY/this.zoomStopHeight);
this.map.zoomTo(zoomLevel);

Replying to elemoine:

* the getResolutionForZoom method of the Map class checks whether that a method of the same name is defined on the base layer. Since the getResolutionForZoom method is part of the generic Layer class, I don't see how this condition wouldn't be met.

Yeah, was thinking a layer might set it to null to exclude itself from this fractional business. Thanks for the tip, I've removed it.

* I'm also not a big fan of the RESOLUTIONS check in allowFracZoom. Why not define the new property allowFracZoom in the Layer class and have the FixedZoomLevels class set this property to false? In that way, layers that don't permit fractional zoom can advertize it - the TileCache layer might be of those classes by the way, and this isn't caught by your RESOLUTIONS check (I'm not too familiar with the TileCache layer and be cautious with my assessments :-)

Gone now. The fractionalZoom property is now a "user beware" property. Works for layers where non-discrete zoom levels makes sense. Doesn't otherwise.

* Line 820 of the patch, you use the i variable while it was defined outside the current scope in the for loop above.

Can you restate this? The patched code does not use i while it is undefined, nor are there scope issues as far as I can tell. I did move the zoom = Math.max(0, i-1) statement into an else block, but i is always defined there. I'm updating the patch. Let me know if you still see issues with it.

allow for non discrete zoom levels

Tests pass in FF and IE7. Thanks for any review.

Tim, I also do not see anything wrong with using the i outside the loop. Furthermore, I have applied your patch and modified my local MapBuilder copy to make use of fractionalZoom. I tested with all MapBuilder examples, and it just works.

The code is non-invasive and intuitive to read. With other words: great work! My vote on this ticket is "Please commit", but I leave the "Review" state untouched, because there is promised feedback pending from pspencer/madair.

I don't think we need to wait for Paul/Mike on this: 3 weeks is long enough to give something a try. (We can always update things before the 2.6 release anyway.) Tim, this looks good; go ahead with it as is, in my opinion.

Replying to ahocevar:

Tim, I also do not see anything wrong with using the i outside the loop.

---

I doubled check and there's indeed nothing wrong with using the i outside the loop.

-- Eric

Yeah, don't wait on us. I've looked this patch over a couple of times but I don't think that Mike has had a chance to try it out.

(In [5982]) Adding fractionalZoom property to the map. This allows zooming to an arbitrary level, making it possible to have non-discrete resolutions for layers that support it. This property should not be set to true for layers with fixed zoom levels (commercial layers or others with cached tiles). r=elemoine,crschmidt,ahocevar (closes #1243)