Latest Contributions by JoelBennett

I too have had the same issue, which has been occurring since geometryEngine's introduction in 3.13.  I hadn't noticed (and therefore can't verify) the try/catch issue since my code doesn't have any of those clauses around this operation.  My code executes the combine operation inside of a callback from a QueryTask, so there's probably a try/catch somewhere up the call stack in the JavaScript API. Since I'm not the only person this is happening to, I was emboldened to look into this further.  Unfortunately, the code for geometryEngine is much too obfuscated to readily understand what's going on, but here's the function where things foul up: b.prototype.iL=function(a){ for(var d=[],c=[],h=this.sn,l=b.bc(h),g=0;g<l.length;g++){   var m=l ,n=h ;   if(this.Fc||1E4<n.hq&&1<n.Ci.length){    this.bp-=n.Ci.length;    for(this.Qo-=n.Ci.length;0<n.Ci.length;)     d.push(n.Cv().hw),n.OQ();    c.push(m)   } } for(g=0;g<c.length;g++)   delete h[c ]; return d }; The error is thrown when comparing if 1E4 is less than n.hq, and occurs because n is undefined.  If we look further up, we see n was set to the value of h ...that is, an element of some array.  In my case, h was an array with a length of 4, but elements with index 0, 1, and 2 were undefined.  However, this code wrongly assumes all the elements in the array it references are defined...hence the error. I use a locally hosted version of the API, and so I tweaked it a little to check for a valid value of n before proceeding. Therefore, the original code (which you can ctrl-f to find): if(this.Fc||1E4<n.hq&&1<n.Ci.length) became if((n)&&(this.Fc||1E4<n.hq&&1<n.Ci.length)) This appears to have fixed the problem, although I haven't performed any rigorous testing...as such, if you try this, proceed with caution.  Also, if you do, you'll probably need to delete temporary internet files before reloading your app to ensure you get the updated version. ... View more

It's been awhile since you posted this, and you may have already figured it out.  For posterity, though, I believe the problem here is that IE only supports custom cursors in .cur and .ani format (not .png). See also: cursor property (Internet Explorer) ... View more

The getUniqueValueInfo function of the UniqueValueRenderer takes a graphic, examines its attributes, and returns an info object with the symbol.  I've copied that same function and updated it to instead return the index it creates from the graphic's attributes.  If you could get a look at these indexes, this information should tell you what values you need to specify when setting up your renderer.  I haven't tested this, but here it is: var featureLayer = map.getLayer("my_layer_id"); var renderer = featureLayer.renderer; var graphics = featureLayer.graphics; renderer.getUniqueValueInfo2 = function(a) { var b = this.attributeField, d = a.attributes, e, f; this._multiple ? (a = this.attributeField2, e = this.attributeField3, f = [], b && f.push(d), a && f.push(d), e && f.push(d ), b = f.join(this.fieldDelimiter || "")) : b = c.isFunction(b) ? b(a) : d; return b; }; for (var x = 0; x < graphics.length; x++) console.info(x.toString() + ": [" + renderer.getUniqueValueInfo2(graphics ) + "]"); ... View more

I'm not sure exactly what's going on, because I have an app that uses esriRequest over SSL and it works ok.  I do have two ideas though: Assuming you have IE 11, you can go to the developer tools (F12), go to the network tab, and do a network trace.  You can examine the requests and responses, and maybe compare what they look like when going directly versus going through esriRequest.  Pretty much all the other browsers, such as Chrome also have this same capability, but I'm most familiar with IE. Also, I don't access the ArcGIS Server directly (through ports 6080 and 6443) - I use the web adaptor for IIS instead.  If you're able, try installing the web adaptor and then take the port numbers out of your urls.  It's a long shot, but it may be that the different port numbers are causing your browser to think that you're making cross-domain requests. ... View more

​Have you configured your ArcGIS Server instance to be accessed over both http and https?  I think the admin console where you do this is at http://server:6080/arcgis/admin if so, what happens when you go directly to the url for the print task? ... View more

Set the wkid of your spatial reference to 4326 instead of 102100.  Your coordinates are in WGS 84, not Web Mercator. ... View more

I'm not...window.dojoConfig.parseOnLoad is explicitly set to false in my application. My Measurement widget gets created programmatically and then added to the DOM tree as a child of another dijit (a subclass of dijit/Dialog that's been hacked to be modeless).  Perhaps adding it as a child of another pre-existing dijit already in the tree makes the difference... ... View more

I can't say with absolute certainty, but I know it worked for me (and now for Arjun as well).  I think this behavior started in 3.10.  I'm supposing it's called automatically and doesn't need to be explicitly called, but I don't see evidence of that in the Measurement module itself... ... View more

I've run into this as well...our map should default to one of several sites, but no amount of fooling with the fitExtent and extent parameters (including shrinking the default extent) produces a sure fit for all of them.  More often, the map starts up one level too far in for some, or one level too far out for others.  And there's also cases where the map should start up where the user previously left off, and even with the same screen size as before, the map still often starts at the wrong zoom level.  It's been giving me fits for awhile, but I've gotten to the bottom of it. The reason this occurs is found in the _coremap module.  In the _addLayerHandler function, observe what happens when the map's graphics container is added (when the map loads): if (a === this.graphics) {   ...   this._firstLayerId = null;   ...   d = this._fixExtent(b, i.fitExtent);   this.extent = d.extent; //THIS LINE CAUSES THE PROBLEM   this.__LOD = d.lod;   this.__setExtent(this.extent, null, null, i.fitExtent);   ... } Notice the call to __setExtent.  One of the first things that happens in __setExtent is another call to _fixExtent - with the already-fixed extent.  This extra call can result in an expanded extent that causes the map to start up one zoom level out from where it should. The simplest fix is to rewrite the problem line to:   this.extent = this.extent || d.extent; If you have a local copy of the API, you can implement this fix on your own, but that can be tricky.  You'd have to update init.js, and finding that particular line can be difficult, especially since variable names can be different. There are workarounds though.  I've written a function called setDefaultExtentOptions that adjusts your default extent to match the actual extent your map will show, and appends appropriate values to the options parameter (the second parameter) that you pass into the map's constructor. The function and workflow would go like: function setDefaultExtentOptions(defaultExtent, mapContainerID, options) { var contentBox = domGeom.getContentBox(document.getElementById(mapContainerID)); var extentHeight = defaultExtent.getHeight(); var extentWidth = defaultExtent.getWidth(); var viewHeight, viewWidth; var lods = options.lods; var dY, dX; var center; var lod; options.spatialReference = defaultExtent.spatialReference; options.extent = defaultExtent; options.fitExtent = true; for (var x = lods.length - 1; x >= 0; x--) {   lod = lods ;   viewHeight = lod.resolution * contentBox.h;   viewWidth = lod.resolution * contentBox.w;   if ((viewHeight > extentHeight) && (viewWidth > extentWidth)) {    center = defaultExtent.getCenter();    dY = viewHeight * 0.5;    dX = viewWidth * 0.5;    options.extent = new Extent(center.x - dX, center.y - dY, center.x + dX, center.y + dY, defaultExtent.spatialReference);    options.fitExtent = false;    options.scale = lod.scale;    options.zoom = lod.level;    options.center = center;    return;   } } } var defaultExtent = new Extent(minX, minY, maxX, maxY, new SpatialReference(wkid)); var divId = "map"; //set according to your div's ID. var options = { //set map options here...MUST INLCUDE VALUE FOR lods }; setDefaultExtentOptions(defaultExtent, divId, options); var map = new Map(divId, options); I implemented this workaround before actually figuring the problem out.  There are probably better workarounds, but optimally, a workaround wouldn't be necessary. I see the question was initially raised for 2.8, but it's still happening in 3.13, which I'm using. ... View more