[postgis-devel] GiST estimate and index_extent()

[strk at refractions.net]

Mark,
Paul actually did not express his preference about this, and
actually I'm not convinced as well...
I'm not sure 2 days will be enough, not for the implementation
but for the decisions reguarding backward-compatibility.
If we find a clean solution today I think that would be ok.

We currently have a fix_geometry_columns() which fixes
f_table_schema, attrelid and varattnum.
The function is automatically invoked by AddGeometryColumn()
and update_geometry_stats().

What we could do is drop the automatic invokation and improve
it to drop attrelid, varattnum and stats if present for PG>=800
and to change stats into a stakind1 smalint / stanumbers1 float[]
for PG<800 if stats is present. Note that in this case the HISTOGRAM2D
type would be pretty useless, so we could drop it as well.

About TODOs, running regression tests I found that lwgparser
does not check polygon validity as did HWGEOM do. Maybe that
should be added, but I guess could wait for 1.0 or RC2.

Comments ?

On Thu, Jan 13, 2005 at 10:34:38AM -0000, Mark Cave-Ayland wrote:
> 
> > -----Original Message-----
> > From: strk at refractions.net [mailto:strk at refractions.net] 
> > Sent: 13 January 2005 08:05
> > To: Mark Cave-Ayland
> > Cc: 'PostGIS Development Discussion'
> > Subject: Re: [postgis-devel] GiST estimate and index_extent()
> > 
> > 
> > Mark, I think RC1 must really come out now...
> > And the histogram2d handling seems not straihtforward to me.
> > 
> > I think what we need is a "live" upgrade script, able to 
> > update a live installation from RC1 up. This would simplify 
> > operations like changing geometry_columns definition and the like.
> > 
> > --strk;
> 
> 
> Hi strk,
> 
> If you really feel that RC1 should come out now then I guess we should wait
> until PostGIS 1.1 as there is no point putting in such a large change after
> going to RC. I appreciate that it is not a simple change to do (it may take
> a couple of days), but I really think it will help debug and problems that
> may occur with the statistics and help simplify the code at the same time.
> 
> It would be have been nice to include this in 1.0 since as it is so
> different to 0.9, I think users would be more tolerant to this type of
> change than they would if, say going from 0.8 to 0.9, or 1.0 to 1.1 - but I
> respect that the final decision rests with you and Paul.
> 
> (BTW this was the only outstanding issue on my TODO list - I guess everyone
> else's TODO list is now empty?)
> 
> 
> Kind regards,
> 
> Mark.
> 
> ------------------------
> WebBased Ltd
> South West Technology Centre
> Tamar Science Park
> Plymouth
> PL6 8BT 
> 
> T: +44 (0)1752 791021
> F: +44 (0)1752 791023
> W: http://www.webbased.co.uk
> 
> 
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel

-- 

For standing up against patentability of software,

  Thank You, Poland!

Read the intervention:    http://kwiki.ffii.org/ConsPolon041221En
Send your thanks:         thankyoupoland.info
Read/do more:		  http://www.noepatents.org/