On May 27, 2011, at 10:33 AM, Flo Ledermann wrote:

> Hi Dane,
> 
> On 05/25/2011 10:12 PM, Dane Springmeyer wrote:
>> re: contributing to the wiki - I would love help. I figured 
>> adding/refactoring this page would be the place to start:
>> 
>> http://trac.mapnik.org/wiki/Troubleshooting
> 
> Funny that you point me to that page, because the issue described on
> that page was another one I was running into (polygons across the
> dateline).

Common story with Natural Earth - rich data but understandably not perfect for 
reprojection.

> But in my case I tried to use the +pm parameter of the PROJ.4
> projection string to "rotate" the earth to a different center meridian
> on purpose.

You are a proj4 wizard I see :)

> 
> In this case neither clipping nor the +over parameter would help me,
> because with the first I would loose the geometry "coming out" on the
> other side, while with the second the whole projection would just be
> shifted and I would end up with a distorted projection that still has
> good ol' Greenwich at the center.

Ya, I would recommend clipping the data rather than using proj4 to adapt. 

> 
> I once wrote a simple map renderer in Java where I included a heuristic
> algorithm to split "wrapping" polygons in two on the fly while rendering
> -- detect the dateline crossover in the polygon's vertices, and in case
> it happens create a second polygon on the fly and assign subsequent
> vertices to it until it crosses back again.
> 
> Would you be in favour of such a feature ever being introduced in
> Mapnik?

If clipping the data ahead of time fixes your problem, that seems like the best 
recommended route. If not, then let's discuss more.


> I am not sure what the performance implications for the 99.9% of
> the cases that don't need that feature would be (I guess it should only
> be an optional rendering mode activated by a flag or something). If
> there are no fundamental objections against it, I could try to reproduce
> what I did in Java in the C++ code (given my coding skills are sufficient)
> 
> Regarding the wiki, are you distinguishing between Mapnik and Mapnik2
> (as there are maybe different solutions available, or even different
> problems), or is this just a temporary distinction?

Use your best judgement - thanks!

> 
> Best,
> 
> Flo
> 

_______________________________________________
Mapnik-users mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/mapnik-users

Reply via email to