[gdal-dev] Ogrtindex: schema consistency check too strict?

[Rahkonen Jukka (MML)]

Hi,

I created ogrtileindex with -accept_different_schemas option and at least I can render WMS maps with Mapserver 6.4 with it. I need those integer attributes for classification but not for WFS at least by now.

I fear that there is no GDAL tool that could unify the data types by lifting Number(9.0) into Number(10.0) in the dfb headers of my shapefiles. Rewriting everything into common schema could be on option even it would take some time because this data set is about 600 GB. But how could I define such a  common output schema?
 
-Jukka Rahkonen-

-----Alkuperäinen viesti-----
Lähettäjä: Even Rouault [mailto:even.rouault at spatialys.com] 
Lähetetty: 17. maaliskuuta 2015 17:52
Vastaanottaja: Rahkonen Jukka (MML)
Kopio: gdal-dev at lists.osgeo.org
Aihe: Re: [gdal-dev] Ogrtindex: schema consistency check too strict?

Jukka,

Accepting both Integer and Integer64 as the same could cause problems. For example let's say you have a ogr tile index whose first dataset has a "foo(Integer)" field and the second dataset is "foo(Integer64)". In gml_types=auto mode, I believe that MapServer will use the schema of the first dataset for the whole tiles, so you could have truncation of Integer64 values.

By the way, Integer64 bit support in MapServer is sitting in this pull request:
https://github.com/mapserver/mapserver/pull/5061

Even

> Hi,
> Probably because of work done for 64 bit integer support ogrtindex in 
> now nagging that schemata of shapefiles with fields defined as 
> Number(9.0) and
> Number(10.0) do not match. The first is reported as Integer and the 
> latter as Integer64. However, that difference should not have any 
> meaning for ogrtindex utility or Mapserver that is using the index file.
>
> For being able to create indexes for our data I have had to run 
> orgtindex with -accept_different_schemas option which is all right 
> otherwise but it accepts also all the meaningful differences which can 
> make real harm for Mapserver.
>
> I think that Integer and Integer64 could be made to mean the same for 
> ogrtindex but I am not sure if it could have some consequences.
>
> -Jukka Rahkonen-
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/gdal-dev
>


--
Spatialys - Geospatial professional services http://www.spatialys.com