KML output

[Steve Lime]

Hi guys: I have been thinking about KML a lot recently. My latest thoughts were that
it would require a native driver (as opposed to working through OGR) to access styling
information and transformation. I was thinking specifically of using libxml and honoring
the gml transformations (e.g. gml_include) but with a kml prefix (e.g. kml_include). It's
likely that users would want to hide, rename or otherwise transform items as well. XSLT
might be another option but would be complex for users. For input I was thinking of 
query result sets as opposed to "map" requests although either could be done since a map
draw is really just a bbox query. I wasn't aware of the Google C++ library was imminent 
and would defer to that instead of libxml (probably would still need metadata driven 
transformations).

The project does look really cool and I'd love to see functionality like this in MapServer via
the CGI, WFS/WMS and MapScript. Like Thomas I would much rather see a C or C++ 
implentation for maximum utility in those other areas.

Steve

>>> thomas bonfort <thomas.bonfort at GMAIL.COM> 09/21/07 5:01 AM >>>
alessandro,

On 9/21/07, Alessandro Pasotti <ale.pas at tiscali.it> wrote:

> The project is GPL and now available for free (as in "beer").

if it's gpl, then that would also mean "as in speech" no?

> http://www.itopen.it/soluzioni/kml-map-server/

I saw your webpage and it looks pretty cool !

>
> Before putting further efforts in this project (that have many performance
> limitations with huge vector layers) an perhaps porting it to Mapserver 5, I
> wish to know if anybody is working in the implementation of this format
> directly in mapserver.

not to my knowledge. but given that google seems to be releasing a c++
kml library in the near future, I would suspect that getting mapserver
to use it isn't too far fetched if the license is compatible. I
suspect there wouldn't be many objections if you wanted to tackle that
;)

>
> I saw a few messages in this mailing list archives but it seems like nobody is
> now working on this.
>
> Also, do you think that a php mapscript solution make sense in the long run?

IMHO, having the support in the core makes more sense, as it opens up
many more usages

>
> BTW, I've tested MS 5 with AGG and it is really wonderful, you've done a very
> good job!

thanks!

tb