[GeoJSON] Fwd: [Geojson] I-D Action: draft-ietf-geojson-04.txt

Sean Gillies sean.gillies at gmail.com
Sat Jun 25 08:54:48 PDT 2016


On Thu, Jun 23, 2016 at 10:39 AM, Jukka Rahkonen <
jukka.rahkonen at latuviitta.fi> wrote:

> Hi,
>
> It looks that with this version the use of any other CRS than CRS:84 is
> denied. If there still happens to be users who prefer some other CRS for
> example because round-trip from the native CRS used in the database into
> CRS:84 and back may change the cooordinates, should they start calling such
> geospatial JSON with some other name than GeoJSON?
>
> Regards,
>
> -Jukka Rahkonen-


Hi Jukka,

What to call the legacy format and the modern format, what names to use,
this is without a doubt a big deal. I think we'll be in good shape with the
publication of this draft, better than we've ever been.

Until now, there has never been a standard for how to ask for GeoJSON by
name. Requests like "can you email me GeoJSON for European country
boundaries?" or "curl https://example.com/data/countries?f=geojson" have
been relying on folk or commonsense semantics of "GeoJSON", not any
standard. Today we can be very specific in an agreed upon, standard manner
and ask "can you email me application/geo+json GeoJSON for European country
boundaries?" or request "curl -H "Accept: application/geo+json"
https://example.com/data/countries" or in a contract write "deliverables
shall use RFC NNNN GeoJSON." I'm going to work hard to make this
specificity normal in the industry.

Should users who want to stay with features of the legacy format change the
name they use? It will certainly help if they do. In future contracts or
documentation, they should write "deliverables shall use geojson.org 2008
GeoJSON with CRS X." Changing names in deployed software is a different
problem. The good news is that GDAL's "GeoJSON" driver will read
application/geo+json GeoJSON with no problems. GDAL only writes legacy 2008
GeoJSON, of course, but in practice future applications will be strongly
motivated to support legacy GeoJSON that uses CRS:84.

(Aside: I'd love to see GDAL get a specific application/geo+json format
driver. The existing GeoJSON format driver conflates legacy 2008 GeoJSON
with other, different JSON formats.)

Web services that return legacy 2008 GeoJSON should never use
"Content-type: application/geo+json" in response headers. This is plainly
wrong. The proper content type to use for legacy 2008 GeoJSON was never
standardized. Many providers used "application/json", which remains a good
option and is distinct from "application/geo+json".

If you call for "GeoJSON" (which never unambiguously named a content type
in a technical sense) without being more specific, you should expect to get
either the legacy or the modern format and should expect to sort out which
one it is by searching for "crs" members.

Yours,

-- 
Sean Gillies
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geojson.org/pipermail/geojson-geojson.org/attachments/20160625/fe76bb76/attachment.htm>


More information about the GeoJSON mailing list