[postgis-devel] [PostGIS] #230: st_expand seems to affect the execution order wich affects st_dwithin

Kevin Neufeld kneufeld at refractions.net
Wed Aug 5 08:27:26 PDT 2009


ST_Distance is more intensive than the operators you mean.  It wouldn't be more intensive than ST_Intersects

I still think your torture tests could be modified to automatically compute a cost per function.  Then we would have 
costs more accurately assigned than simply blanketing everything with the same cost.

http://postgis.refractions.net/pipermail/postgis-devel/2009-July/006109.html

Just a thought,
Kevin

PostGIS wrote:
> #230: st_expand seems to affect the execution order wich affects st_dwithin
> ----------------------+-----------------------------------------------------
>   Reporter:  nicklas  |       Owner:  robe                     
>       Type:  defect   |      Status:  assigned                 
>   Priority:  medium   |   Milestone:  postgis 1.4.1            
>  Component:  postgis  |     Version:  1.4                      
> Resolution:           |    Keywords:  st_dwithin st_expand bbox
> ----------------------+-----------------------------------------------------
> Comment (by robe):
> 
>  Well the main issue is that costs were introduced in 8.3 so they aren't
>  supported in 8.2.  That is one reason we haven't applied costs.  So we'll
>  need ot incorporate it in such as fashion as to not apply to 8.2.
> 
>  The other issue is that we just haven't gotten around to applying costs to
>  all the functions.  I think we can take an easy rule of thumb and just set
>  them all to 100 except for the operators though.  Then apply say 200 for
>  more intensive like ST_Distance.
> 



More information about the postgis-devel mailing list