[OSM-dev] nominatim / windows

[Sarah Hoffmann]

On Tue, Jan 21, 2014 at 11:23:58AM +0100, Dominik Perpeet wrote:
> Thanks for the pointers to the relevant code sections. I don't think the
> error is there, though:
> 
> - The script doesn't seem to abort, since in the code pretty much the
> last action is to create the table wikipedia_redirect, which was indeed
> created in my setup.
> - placex and place have their triggers and are not empty

That's good then. There was no CREATE TRIGGER message in your output.

What should happen is that the data gets copied from place to placex.
The trigger of placex computes some additional values for the objects
and, most importantly, sets the value of indexed_status to 1. Only the
rows that have this indexed status will later be reindexed (the step
which reports processing 0 objects in your logs).

> Some other step seems to fail silently, any ideas?

The placex table looks a bit small. I'd expect the number of rows to be in
the same order of magnitude as place. So I wonder if the data loading
happens at all. There is certainly no mentioning of it in the logs.
Try to list the row counts by class/type column.
The numbers should be mostly the same for place and placex. If placex
only contains postcodes and some state boundaries (both are imported
from a different source than the place table)	then the data loading
surely failed.

Sarah


> > Looking at the table sizes, it looks like loading the data[1] does not work
> > properly. There should be some INSERT triggers on placex which seem to be
> > missing. They should be set from an SQL script around here[2].
> >
> > My guess would be that your version of psql immediately stops the scripts
> > when it encounters an error while the unix version just skips over them.
> > There are some expected fails which you can also see in your logs:
> >
> > <snip>
> > FEHLER:  Typ A»place_boundingboxA« existiert nicht
> > FEHLER:  Typ A»place_boundingboxA« existiert nicht
> > FEHLER:  Typ A»place_boundingboxA« existiert nicht
> > LINE 4:   result place_boundingbox;
> > <snap>
> >
> > It should really just ignore them and continue.
> >
> > Sarah
> >
> >
> > [1] https://github.com/twain47/Nominatim/blob/master/utils/setup.php#L377
> > [2] https://github.com/twain47/Nominatim/blob/master/sql/tables.sql#L224
>