[OpenLayers-Trac] Re: [OpenLayers] #3030: ArcGIS (AGS) Tile Cache Layer Contribution

[OpenLayers]

#3030: ArcGIS (AGS) Tile Cache Layer Contribution
--------------------------------------+-------------------------------------
 Reporter:  dmiddlecamp               |       Owner:  euzuro         
     Type:  feature                   |      Status:  new            
 Priority:  minor                     |   Milestone:  2.11 Release   
Component:  Layer                     |     Version:  2.10           
 Keywords:  ags, esri, layer, review  |       State:  Needs More Work
--------------------------------------+-------------------------------------

Comment(by dmiddlecamp):

 Thanks again for your efforts moving this functionality forward.  I really
 appreciate your feedback, and I wanted to try and address your changes /
 questions:

  * getMaxExtent override:
    * Cool, probably a change I wouldn't have considered since it involves
 changing core functionality beyond this layer, but it seems like a clever
 solution to the larger issue of changing extents.
  * use zeroPad from tschaub's sandbox instead
    * I had missed this function, which does largely the same as that small
 helper 'padLeft' that I included.  The major difference is that 'padLeft'
 lets you specify the pad char, and doesn't require the initial value to be
 numeric.  Tschaub's is more specific to this situation, and that probably
 makes calls to it a bit cleaner.
  * inherit from !OpenLayers.Layer.XYZ instead
    * XYZ and TMS seem fairly similar to me with the exception of the
 maxExtent / tileOrigin difference.  I had the impression that the ArcGis
 cache layers made use of the tileOrigin as well as the maximum extent, so
 I attempted to pick the layer closest to the server behavior.  XYZ also
 seems to include some auto-configuration behavior for sphericalMercator
 that doesn't seem to match the server-specific initialization concept of
 the ArcGISCache layer.  TMS seemed the most appropriate, and it's been
 working well so far, but if someone with more experience recommends XYZ,
 and it maintains the desired behavior, then I'm all for it.
  * Why do you add 0 to the lod.scale in initialize?
    * That is probably leftover debug code, amazing those 3 characters
 stayed in there for so long.  :)
  * What is the main difference between your getURL function and the one
 from tschaub in his sandbox? Any chance you can adapt his work instead of
 rewriting? Or at least point out why you did it differently.
    * We had a version of this layer internally that I think may have been
 based on tschaub's initial code as well as that of my peers here. I was
 originally trying to debug that version, and fix the issues with it across
 the different tile and server configurations we needed to support.  It
 made sense to develop against that layer instead of another user's
 sandbox, and through the course of debugging, I was able to add code to
 help bridge the differences between the Esri server capabilities object
 and OpenLayers.  In the event that some small math transcription was the
 source of the problem, a fresh code slate seemed safer.  Also, I suspected
 it might not be well received to start suggesting patches to someone
 else's sandbox... :)

-- 
Ticket URL: <http://trac.openlayers.org/ticket/3030#comment:10>
OpenLayers <http://openlayers.org/>
A free AJAX map viewer