[gdal-dev] S-57 Driver WRONG Geometry

[ogr user]

Correct me if I am wrong (I might be) but my reading of S-57 standard
(section 4.7 and thereabouts) is such that it does not provide for
multi-line geometries. Instead, it is suggested that if certain edges of
a line have to be "invisible" they should be masked by using appropriate
attribute. I have to admit that I haven't seen that used in major data
sets (NOAA, UK, various european) either. In any case - while it is
certainly possible to store multiple "not really connected" edges for a
given feature, this does not seem to be something that S-57 standard
permits or provides for.

By the same token, standard is being quite specific about not permitting
multiple polygons in the single area feature (though they do provide for
"holes" in a given polygon).

 FWIW I would consider an S-57 ENC using such line structure to be not
well formed. Just my 5 cents.
--Mike

Frank Warmerdam wrote:
> Bogdan Grama wrote:
>> Dear Frank
>>
>> Please find in the archive (http://www.wgs.ro/gdal/0_w.zip):
>>
>> -the ENC
>> -GDAL Interpretation
>> -correct  Interpretation
>>
>> The assembled geometry is not according with specifications. Each
>> referenced edge starts and ends with a ConnectedNode. Not all the
>> Connected nodes are present in the feature geometry dump.
>>
>> This is not all. If the file is rendered with Sevencs
>> (http://www.sevencs.com/service-products/download) and few others
>> viewers the rendering is correct.
>> For GDAL, ESRI FME, Caris the geometry is not correct.
>>
>
> Folks,
>
> I have determined this is the same as the previously reported:
>
>   http://trac.osgeo.org/gdal/ticket/2147
>
> for which I neglected to apply the patch.  I have revised the patch
> and applied it in trunk and 1.7 branch.
>
> Best regards,