[postgis-devel] 1, dot, 3 dot 4!

Obe, Regina robe.dnd at cityofboston.gov
Fri Oct 17 08:59:16 PDT 2008


Regina,

> What are the contents of the KML regression diff?
I'll get back to you on this one.  Don't have the box here.



> What does ST_HasBBOX say about the geometries in your two tables? The
> only thing I did that could (in my mind) slow things down for an
> ST_DWithin() test is the index detoasting change.

They all have BBOXes
e.g

SELECt count(p.pid), COUNT(CASE WHEN ST_HasBBOX(p.the_geom) THEN 1 ELSE
NULL END)
FROM landparcels p INNER JOIN neighborhoods n ON ST_DWithin(p.the_geom,
n.the_geom, 200)
WHERE n.name = 'Mattapan'

returns 7873;7873 .  I checked my neighborhoods table and they all have
bboxes too.

I'm still leaning toward the side benefit of ST_Distance caching that is
lost in the new implementation, because my timings often get better with
each increased buffer in the old install, but don't have as much of an
effect if any in new install.  Things seem to cache very quickly so its
hard to run the same tests.  You wouldn't happen to know by chance how
to clear pg cache buffer?

For example - running these queries in sequence - old implementation
does much better which unfortunately for me is a very common use case
since people do all sorts of what if analysis like (what if I increased
the buffer x until we have at least 1000 residents to target)

SELECT count(p.pid), COUNT(CASE WHEN ST_HasBBOX(p.the_geom) THEN 1 ELSE
NULL END)
FROM landparcels p INNER JOIN neighborhoods n ON ST_DWithin(p.the_geom,
n.the_geom, 200)
WHERE n.name = 'Mattapan'

SELECT count(p.pid), COUNT(CASE WHEN ST_HasBBOX(p.the_geom) THEN 1 ELSE
NULL END)
FROM landparcels p INNER JOIN neighborhoods n ON ST_DWithin(p.the_geom,
n.the_geom, 300)
WHERE n.name = 'Mattapan'

SELECT count(p.pid), COUNT(CASE WHEN ST_HasBBOX(p.the_geom) THEN 1 ELSE
NULL END)
FROM landparcels p INNER JOIN neighborhoods n ON ST_DWithin(p.the_geom,
n.the_geom, 400)
WHERE n.name = 'Mattapan'


> On another note - I still would really like to wait till we release
> 3.1.0 and Mark can include that in the Windows build instead of 3.0.1.
> It seems a shame to let all that hard-work go to waste.

> It could be quite a while before 3.1.0 can get out. I labor under my
> lack of C++ and noone else is stepping up. Things like removing
> warnings and so on.

Alright a bit of a bummer - oh well.  As Chris H said before - the
weakness of the flesh
is upon me. My spirit is strong, but the flesh is weak :(  Well maybe it
will get stronger some day.  So far you've been doing pretty good though
:).

Thanks,
Regina
-----------------------------------------
The substance of this message, including any attachments, may be
confidential, legally privileged and/or exempt from disclosure
pursuant to Massachusetts law. It is intended
solely for the addressee. If you received this in error, please
contact the sender and delete the material from any computer.



More information about the postgis-devel mailing list