[jts-devel] A comment to JTS Technical Specification

[Martin Davis]

Darn - I guess we guessed wrong.  (Actually, it makes sense that spec takes precedence over the examples, but my main motivator for the choice was looking at what other implementations did - and they got it wrong too, apparently).

I guess the best solution is to enhance the reader to accept either form, and modify the writer to output the correct syntax.

> What does this adoption mean?

It means JTS currently uses the format with fewer brackets.

Martin Davis, Senior Technical Architect
Vivid Solutions Inc.      www.vividsolutions.com
Suite #1A-2328 Government Street Victoria, B.C. V8T 5G5
Phone: (250) 385 6040 - Local 308 Fax: (250) 385 6046


> -----Original Message-----
> From: jts-devel-bounces at lists.jump-project.org 
> [mailto:jts-devel-bounces at lists.jump-project.org] On Behalf 
> Of Mateusz Loskot
> Sent: April 17, 2006 1:41 PM
> To: jts-devel at lists.jump-project.org
> Subject: [jts-devel] A comment to JTS Technical Specification
> 
> 
> Hi,
> 
> Recently, I was reading JTS Technical Specification and I 
> encountered some interesting note about inconsistency in 
> Well-Known-Text format definition.
> 
> The specification says:
> 
> "Note that there is an inconsistency in the SFS. The WKT 
> grammar states that MultiPoints are represented by 
> "MULTIPOINT ( ( x y), (x y) )", but the examples show 
> MultiPoints as "MULTIPOINT ( x y, x y )". Other 
> implementations follow the latter syntax, so JTS will adopt 
> it as well."
> 
> I tried to find some update regarding this issue and there is 
> a note on OpenGIS forum. Here is how OGC experts explain it:
> 
> (Source:
> http://feature.opengeospatial.org/forumbb/viewtopic.php?t=301)
> 
> //////////////////////////////////////////////////////////////
> ///////////
> // Date: Feb 27, 2006
> 
> "He is right on the error, but wrong on the correction. 
> According to ISO rules, the constructions (normative text) 
> take precedence over examples (informative text) and 
> therefore the example is wrong, and should be:
> 
> MULTIPOINT((10 10), (20 20))
> 
> This is more consistent with the other GeomCollection such as:
> 
> MULTILINESTRING ((10 10, 20 20), (15 15, 30 15))
> 
> And (the extra "()" is needed to group rings into polygons):
> 
> MULTIPOLYGON (((10 10, 10 20, 20 20, 20 15, 10 10)),((60 60, 
> 70 70, 80 60, 60 60 )))
> 
> There probably is a good argument for using the 
> GeomCollection pattern and requiring tagged text at each point
> 
> GEOMETRYCOLLECTION(POINT (10 10), POINT (30 30), LINESTRING 
> (15 15, 20 20))
> 
> Which would make it:
> 
> MULTIPOINT(POINT(10 10), POINT(20 20))
> 
> but that would have to be change requested since it modifies 
> normative text.
> 
> // EOF 
> //////////////////////////////////////////////////////////////
> ///////////
> 
> So, as you can read in the note above, examples should be 
> considered as "wrong".
> 
> I'd like to ask how JTS handles this issue in details.
> How should I understand this statement?
> 
> "Other implementations follow the latter syntax, so JTS will 
> adopt it as well."
> 
> What does this adoption mean?
> 
> Thanks for your help in advance
> 
> Cheers
> -- 
> Mateusz Łoskot
> http://mateusz.loskot.net
> 
> _______________________________________________
> jts-devel mailing list
> jts-devel at lists.jump-project.org 
> http://lists.refractions.net/mailman/listinfo/jts-devel
>