Re: [sqlite] False negatives from RTree

2011-07-05 Thread Ben Harper
I've tried stepping into the rtree code, and the right code paths
all seem to get hit.
Nevertheless, once the transaction completes, I open the table again,
and the freshly inserted records do not appear in the 
idx_elec_lv_cable_geometry_rowid table. And this is especially strange,
since I debug-step-over the code that performs the insertion into
idx_elec_lv_cable_geometry_rowid, and it all appears to run without
a hitch.



-Original Message-
From: sqlite-users-boun...@sqlite.org [mailto:sqlite-users-boun...@sqlite.org] 
On Behalf Of Richard Hipp
Sent: 05 July 2011 03:20 PM
To: General Discussion of SQLite Database
Subject: Re: [sqlite] False negatives from RTree

On Tue, Jul 5, 2011 at 8:55 AM, Ben Harper <b...@imqs.co.za> wrote:

> Hi,
> Has anybody ever seen anomalous behaviour on an R*Tree. Specifically, false
> negatives?
> I'm using Spatialite 2.4.0 RC.
>
> What I end up with is a table into which I can insert new geometry, but
> that geometry does not get retrieved by an appropriate RTree search.
> Indeed, even a search which spans all of my geometry does not yield these
> newly inserted records.
>
> For example:
> select count(rowid) from idx_elec_lv_cable_geometry where xmax >= -9e30 AND
> xmin <= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;
> 17235
>
> select (count*) from elec_lv_cable;
> 17241
>
> So I'm missing 6 records there.
>

What does the following show:

SELECT * FROM idx_elec_lv_cable_geometry
EXCEPT SELECT * FROM idx_elec_lv_cable_geometry where xmax >= -9e30 AND xmin
<= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;

That query should show just the 6 records that are missing.


>
> I can't discern anything strange about the data, or my methods.
> I'm curious to know whether anybody has seen this kind of thing before?
>
> Thanks,
> Ben
>
> ps. This message is cross-posted to the Spatialite mailing list.
> ___
> sqlite-users mailing list
> sqlite-users@sqlite.org
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
>



-- 
D. Richard Hipp
d...@sqlite.org
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] False negatives from RTree

2011-07-05 Thread Ben Harper
The statement <<

SELECT * FROM idx_elec_lv_cable_geometry
EXCEPT SELECT * FROM idx_elec_lv_cable_geometry where xmax >= -9e30 AND xmin
<= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;

>>

yields an empty set.



-Original Message-
From: sqlite-users-boun...@sqlite.org [mailto:sqlite-users-boun...@sqlite.org] 
On Behalf Of Richard Hipp
Sent: 05 July 2011 03:20 PM
To: General Discussion of SQLite Database
Subject: Re: [sqlite] False negatives from RTree

On Tue, Jul 5, 2011 at 8:55 AM, Ben Harper <b...@imqs.co.za> wrote:

> Hi,
> Has anybody ever seen anomalous behaviour on an R*Tree. Specifically, false
> negatives?
> I'm using Spatialite 2.4.0 RC.
>
> What I end up with is a table into which I can insert new geometry, but
> that geometry does not get retrieved by an appropriate RTree search.
> Indeed, even a search which spans all of my geometry does not yield these
> newly inserted records.
>
> For example:
> select count(rowid) from idx_elec_lv_cable_geometry where xmax >= -9e30 AND
> xmin <= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;
> 17235
>
> select (count*) from elec_lv_cable;
> 17241
>
> So I'm missing 6 records there.
>

What does the following show:

SELECT * FROM idx_elec_lv_cable_geometry
EXCEPT SELECT * FROM idx_elec_lv_cable_geometry where xmax >= -9e30 AND xmin
<= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;

That query should show just the 6 records that are missing.


>
> I can't discern anything strange about the data, or my methods.
> I'm curious to know whether anybody has seen this kind of thing before?
>
> Thanks,
> Ben
>
> ps. This message is cross-posted to the Spatialite mailing list.
> ___
> sqlite-users mailing list
> sqlite-users@sqlite.org
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
>



-- 
D. Richard Hipp
d...@sqlite.org
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users


Re: [sqlite] False negatives from RTree

2011-07-05 Thread Richard Hipp
On Tue, Jul 5, 2011 at 8:55 AM, Ben Harper  wrote:

> Hi,
> Has anybody ever seen anomalous behaviour on an R*Tree. Specifically, false
> negatives?
> I'm using Spatialite 2.4.0 RC.
>
> What I end up with is a table into which I can insert new geometry, but
> that geometry does not get retrieved by an appropriate RTree search.
> Indeed, even a search which spans all of my geometry does not yield these
> newly inserted records.
>
> For example:
> select count(rowid) from idx_elec_lv_cable_geometry where xmax >= -9e30 AND
> xmin <= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;
> 17235
>
> select (count*) from elec_lv_cable;
> 17241
>
> So I'm missing 6 records there.
>

What does the following show:

SELECT * FROM idx_elec_lv_cable_geometry
EXCEPT SELECT * FROM idx_elec_lv_cable_geometry where xmax >= -9e30 AND xmin
<= 9e30 AND ymax >= -9e90 AND ymin <= 9e30;

That query should show just the 6 records that are missing.


>
> I can't discern anything strange about the data, or my methods.
> I'm curious to know whether anybody has seen this kind of thing before?
>
> Thanks,
> Ben
>
> ps. This message is cross-posted to the Spatialite mailing list.
> ___
> sqlite-users mailing list
> sqlite-users@sqlite.org
> http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
>



-- 
D. Richard Hipp
d...@sqlite.org
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users