[postgis-users] Geometry appears empty

Tue Jul 1 22:56:57 PDT 2008

On Tue, 2008-07-01 at 21:53 -0700, geoffi wrote:
> Hi Tim,
> 
> OK to clarify...
>  I have 694 records in my table, all have geometry based on the query:
>     select count(*) from gen_poly
>     where ST_IsValid(geom);
> 
>        count
>        -------
>           694
>       (1 row)

Ok, that's confirmed the geometries are there.

> 
> When viewed in pgAdminIII only a handfull "appear" to have actual data in
> the geom column (see original post). This is also the result when queried
> using psql. The rest "appear" to be null.
> 
psql should be showing the geometries.  What query are you using to see
them?

Using psql try:
SELECT AsText(geom) AS geom from <mytable>;
That should show you the text representations of the geometries and
return 694 rows.

> FME proves that all 694 geometries exist and can display and interrogate
> them.
> 
> Pushing the data out as a WMS via GeoServer only displays the handfull that
> can be "seen" in psql and pgAdminIII.
> 

Are you sure it is *exactly* the same subset that doesn't get shown in
both geoserver and PgAdmin?

Does geoserver log any errors?  Are you having minscale or maxscale
issues in visualising the data?  Given FME can see the data it is in
there and appears to be doing what it is supposed to do. I'm starting to
think it is an issue with geoserver of some sort.

Regards,
Tim Bowden

> So, we've established that all geometries are valid, but only some can be
> seen using the Postgres tools or GeoServer. FME can see ALL the geometries!
> 
> 
> 
> 
> Tim Bowden wrote:
> > 
> > 
> > On Tue, 2008-07-01 at 20:28 -0700, geoffi wrote:
> >> Hi Tim,
> >> 
> >> First of all I can confirm that the geometries are not null by:
> >>    a) select count(*) from <my_table> where geom is null;
> >>        This returns a count of 0.
> >>    b) the FME Universal Viewer successfully displays all the geometries
> >> successfully. Features can be graphically selected and their attributes
> >> interrogated.
> > 
> > Ok, so you're saying there is geometry data is in postgis for each
> > record?  You seem to be negating this further down.
> > 
> >> 
> >> The link you gave seems to imply that pgAdminIII may not be the tool to
> >> use
> >> to verify geometry columns. It also mentions a possible limitation for
> >> the
> >> number of verticies hitting a limit of 1365. I will need to check my
> >> geometries if this is the same in my case.
> >> 
> > 
> > This limitation is just PgAdminIII's ability to view the data.  It does
> > not affect what's in the DB.  I'd be  surprised if there isn't a fix on
> > the way for PgAdminIII as this has been around for a while now.
> > 
> >> In the meantime, I've also used the psql command line interface to check
> >> the
> >> records with the same results as per the pgAdminIII. Can you recommend
> >> another tool I can use to report the geometry data?
> >> 
> > 
> > Prefer psql over PgAdmin myself (can't seem to break the cli habit).
> > There should be no issues with using psql.  Are you saying some records
> > don't contain geometry data (as per original PgAdmin view) or you got no
> > null geom records?  Little confused now.
> > 
> >> As mentioned in the original post, my final step is to publish a WMS via
> >> GeoServer. Only those geometries that "appear" to be populated come
> >> through
> >> in the WMS. While this may be a GeoServer issue, it is still reporting
> >> the
> >> same information that "appears" in PostGIS.
> >> 
> >> Any other thoughts?
> >> 
> > Check the validity of your geoms.
> > select count(*) from <table> where not ST_IsValid(the_geom);
> > (or whatever your geom column is called).
> > 
> > Regards,
> > Tim Bowden
> >> 
> >> 
> >> 
> >> 
> >> Tim Bowden wrote:
> >> > 
> >> > Geoff,
> >> > 
> >> > This earlier thread may be of interest:
> >> >
> >> http://postgis.refractions.net/pipermail/postgis-users/2008-March/018814.html
> >> > 
> >> > It deals with limitations of pgadmin and postgis.
> >> > 
> >> > If the geometries really are null, are you also able to try loading the
> >> > data with ogr2ogr to see if you get the same result?   Are you sure you
> >> > don't have any invalid geometries?  I don't know how FME handles these.
> >> > 
> >> > HTH,
> >> > Tim Bowden
> >> > On Tue, 2008-07-01 at 17:12 -0700, geoffi wrote:
> >> >> I have translated some polygon data into a PostGIS table using FME.
> >> The
> >> >> translation was successful with no errors reported. All the records
> >> >> appear
> >> >> in the PostGIS table (using the pgAdminIII application) except for
> >> about
> >> >> 90%
> >> >> of the geometries (see image below).
> >> >> 
> >> >> http://www.nabble.com/file/p18227858/table.jpg 
> >> >> 
> >> >> However, using the FME Universal Viewer all records are displayed with
> >> >> their
> >> >> correct geometry as I would expect.
> >> >> 
> >> >> What is causing the Geometry column in the table to appear empty?
> >> >> Is there a setting in PostGIS to force the data to be visible in the
> >> >> pgAdminIII application?
> >> >> 
> >> >> Attempts to generate a WMS from this table using GeoServer only
> >> displays
> >> >> those records that appear to have geometry.
> >> >> 
> >> >> Performing a similar exercise using point geometries, there was no
> >> >> problem.
> >> >> 
> >> > -- 
> >> > vim is a good example of an IDE done properly.  emacs is an example of
> >> > an IDE done poorly.  Anything that involves little pictures is a
> >> > mis-umderstanding of what an IDE should be.
> >> > 
> >> > _______________________________________________
> >> > postgis-users mailing list
> >> > postgis-users at postgis.refractions.net
> >> > http://postgis.refractions.net/mailman/listinfo/postgis-users
> >> > 
> >> > 
> >> 
> > -- 
> > vim is a good example of an IDE done properly.  emacs is an example of
> > an IDE done poorly.  Anything that involves little pictures is a
> > mis-umderstanding of what an IDE should be.
> > 
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at postgis.refractions.net
> > http://postgis.refractions.net/mailman/listinfo/postgis-users
> > 
> > 
> 
-- 
vim is a good example of an IDE done properly.  emacs is an example of
an IDE done poorly.  Anything that involves little pictures is a
mis-umderstanding of what an IDE should be.