Re: [Geojson] PR #199: CRS

[Martin Thomson <[e-mail]>]

On 22 April 2016 at 00:58, Tim Schaub <[e-mail]> wrote:
> Once you say that you can
> only understand the axis order if you also have a database of every
> possible CRS, you make lightweight web clients impractical.

That's a very narrow use case.  I find that geographical data tends to
be much less useful without being able to relate it to other
geographical data (such as maps, other features I know about, etc...).

Quoting now from some smart individual on github (not me):

>>>
>From @martinthomson's comment above:

> GeoJSON documents MUST NOT use a coordinate reference system other than the one defined here.

This is such clear, simple language. I think the spec would be
significantly approved if we would use it. I'm wondering what negative
outcomes we might expect if we adopted it.

Many of us use GeoJSON-like serializations in situations like
@sgillies describes above, and I imagine we would continue to write
and/or use lenient parsers to do this. I don't think that these
lenient (or CRS agnostic) parsers would disappear if the spec mandated
that alt-CRS not be used.

Even a strict parser would have to allow a crs member (as a foreign
member), so existing alt-CRS GeoJSON documents would still be valid in
tomorrow's compliant parsers. GeoJSON doesn't specify the range of
allowed numbers for coordinates, so writing a parser that throws in
the face of large/small values would be going beyond what is specified
(and fragile).

Maybe this is discussion for a mailing list, but I'm curious to hear
(again) objections to this language[...]