[GeoJSON] Plug-Pull request to finalize a CRS-free I-D revision

Mike Bostock mike at ocks.org
Thu Jun 13 16:43:51 PDT 2013


>
> IMO, this is because we didn't do it well and good examples never appeared.


I would also speculate that this is partly lack of demand. On the web,
where GeoJSON is the de facto format, web Mercator is the de facto
projection. Most mapping tools are slippy maps, and most of these allow
zooming from whole-world to local-area, so web Mercator is a reasonable
default choice (if not the only sensible choice, given tiles).

But I am hoping in the future (with tools like D3), it will be easier for
people to use projections more appropriate to their applications. Not
everything has to be a slippy map. So perhaps despite limited demand now
for non-Mercator projections, awareness and demand may increase in the
future.

In my workflow, I prefer to only go one way: from spherical coordinates to
projected coordinates. D3 does resampling of spherical polygons [1], so I
consider spherical coordinates the "archival" representation and only
convert to projected coordinates when I want to fix the projection for
display. Converting back to spherical coordinates is possible using the
inverse projection, but I’ve never implemented inverse resampling; I only
do that via ogr2ogr when processing shapefiles.

When handling projected coordinates in the client, I typically don’t need
to know the projection because I’m displaying the coordinates as-is. [2]
This is partly why I’ve never implemented a CRS parser for D3; I’ve never
needed to reproject in the client.

Also, hello everyone. It’s nice to be here, and thank you for the lovely
format. :)

Mike

[1] http://bost.ocks.org/mike/example (part 1)
[2] http://bl.ocks.org/mbostock/5557726
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geojson.org/pipermail/geojson-geojson.org/attachments/20130613/d51642e8/attachment-0005.htm>


More information about the GeoJSON mailing list