[mapserver-dev] Tilecache WMS-C patch for 5.6.5

[Stephen Woodbridge]

Yewondwossen Assefa wrote:
> On 09/08/2010 11:23 AM, Stephen Woodbridge wrote:
>> Hi Assefa,
>>
>> I could not find a bug for this feature, did you ever open one? I will 
>> if you would like. I think it would be a good place to collect 
>> patches, comments and what nots. For example I have patches for 5.4.2, 
>> 5.6.1, 5.6.4, and now 5.6.5.
>>
>> To use this patch, get a clean copy of mapserver-5.6.5, apply tha 
>> patch to it, run autoconf, ./doit-mapserver-svn is my configure 
>> script, make it as normal.
>>
>> I have compiled and linked the above on linux, but have not had time 
>> to do any testing yet, but the changes between 5.6.4 and 5.6.5 do not 
>> really impact the tilecache code. I made changes to the windows build 
>> files, but have no way to test them here.
>>
>  No there is no bug currently on it. Yes please add a bug on it.

I'll create on tomorrow and add the various patches to it.

>> Next I am going to look into adding support for VirtualEarth/Bing 
>> quadtree URLs. I think the tasks related to this are:
>>
>> 1. how to select which tile scheme? I'm thinking it should be defined 
>> in the tilecache.cfg file but could be defined in the mapfile like 
>> map->web->metadata->tilescheme.
>>
>> 2. tools to read, write, validate quadtree URI and map them to tiles 
>> and metatiles
>>
>> 3. algorithm mapping tile to a directory structure. My plan here is to 
>> define MAX_FILES_PER_DIR=1024 or something like that and then have 
>> tiles cached into directories so we have some control over the number 
>> of directories and the number of files in a directory because these 
>> all have performance implications.
>>
>> I would appreciate any ideas, thoughts, collaboration you might be 
>> able to offer.
>>
>  I have not looked into details on the VirtualEarth/Bing quadtree, but 

You can read about it here:
http://msdn.microsoft.com/en-us/library/bb259689.aspx

> one question that comes in mind is there for you  a need to store tiles 
> using the VE structure or are you looking mainly to be able to 
> parse/respond VE type urls that use a quadtree value?

Hmmm, that is a good question. I think the requirement is to be able to 
request tiles based on the quadtree value. So are you suggesting that we 
might be able to easily map the quadtree to the existing tile naming and 
storage mechanism? If so that might greatly reduce the code needed to 
implement support for VE quadtree requests.

-Steve