[GRASS5] v.external FID problem: PostGIS/OGR

GRASS GIS GRASS GIS developer
, ,
1

Hi,

I tried to connect to a PostGIS map which I created last year, but:

v.external dsn="PG:host=myserver user=postgres dbname=spearfish" layer=streams output=pgstreams
ERROR 1: ERROR: Attribute "fid" not found

DBMI-OGR driver error:
Cannot select:
select FID from streams where FID = -1

Building topology ...
WARNING: Random read is not supported by OGR for this layer, cannot build
         support.

Here I found a reference to OGC_FID

http://grass.itc.it/pipermail/grass5/2004-May/014395.html
http://www.gdal.org/ogr/drv_pg.html
"If there is an "ogc_fid" field, it will be used to set the feature
  id of the features"

Obviously there is no ogc_fid column
spearfish=# \d streams;
           Table "public.streams"
Column | Type | Modifiers
--------+-----------------------+-----------
cat | integer |
label | character varying(10) |

but once v.external was functional (maybe dynamic creation of ogc_fid?).

Does anyone have insights if OGR changed?

Thanks

Markus

2

On 5/18/05, Markus Neteler <neteler@itc.it> wrote:

Hi,

I tried to connect to a PostGIS map which I created last year, but:

v.external dsn="PG:host=myserver user=postgres dbname=spearfish" layer=streams output=pgstreams
ERROR 1: ERROR: Attribute "fid" not found

DBMI-OGR driver error:
Cannot select:
select FID from streams where FID = -1

Building topology ...
WARNING: Random read is not supported by OGR for this layer, cannot build
         support.

Markus,

The above warning message is interesting. I assume it is the grass
code that generates this? I wonder how it is determining if OGR
supports random read. It is quite possible that the TestCapabilities
is not returning the correct result if that is being used.

Here I found a reference to OGC_FID

http://grass.itc.it/pipermail/grass5/2004-May/014395.html
http://www.gdal.org/ogr/drv_pg.html
"If there is an "ogc_fid" field, it will be used to set the feature
  id of the features"

Obviously there is no ogc_fid column
spearfish=# \d streams;
           Table "public.streams"
Column | Type | Modifiers
--------+-----------------------+-----------
cat | integer |
label | character varying(10) |

but once v.external was functional (maybe dynamic creation of ogc_fid?).

Does anyone have insights if OGR changed?

I was going to write a big speel about _rowid_ and ogc_fid, but now
realize I am mixing up PostGIS support with SQLite which I was also
working on. OGR might require the OGC_FID to do random reads
(fid based reads), and if the v.support stuff requires that you may well
be out of luck.

I don't think this aspect of OGR changed recently though.

Best regards,
--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam@pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush | Geospatial Programmer for Rent

3

Frank Warmerdam wrote:

On 5/18/05, Markus Neteler <neteler@itc.it> wrote:

Hi,

I tried to connect to a PostGIS map which I created last year, but:

v.external dsn="PG:host=myserver user=postgres dbname=spearfish" layer=streams output=pgstreams
ERROR 1: ERROR: Attribute "fid" not found

DBMI-OGR driver error:
Cannot select:
select FID from streams where FID = -1

Building topology ...
WARNING: Random read is not supported by OGR for this layer, cannot build
        support.

Markus,

The above warning message is interesting. I assume it is the grass
code that generates this? I wonder how it is determining if OGR
supports random read. It is quite possible that the TestCapabilities
is not returning the correct result if that is being used.

Yes, it is using OGR_L_TestCapability, but to work properly the FID should be persistent which probably is not true for PostGIS as mentioned in the link below.

Here I found a reference to OGC_FID

http://grass.itc.it/pipermail/grass5/2004-May/014395.html
http://www.gdal.org/ogr/drv_pg.html
"If there is an "ogc_fid" field, it will be used to set the feature
id of the features"

Obviously there is no ogc_fid column
spearfish=# \d streams;
          Table "public.streams"
Column | Type | Modifiers
--------+-----------------------+-----------
cat | integer |
label | character varying(10) |

but once v.external was functional (maybe dynamic creation of ogc_fid?).

Does anyone have insights if OGR changed?

I was going to write a big speel about _rowid_ and ogc_fid, but now
realize I am mixing up PostGIS support with SQLite which I was also
working on. OGR might require the OGC_FID to do random reads
(fid based reads), and if the v.support stuff requires that you may well
be out of luck.

Would it be possible to support also OID, which is created by default, so many people use it.

Radim

I don't think this aspect of OGR changed recently though.

Best regards,