[postgis-devel] Vive Doublebox -- need to have upgrade scripts fail gracefully at this point

Paragon Corporation lr at pcorp.us
Wed Dec 7 12:45:04 PST 2011


Given that it looks like we'll let this double thing live, we need to make
sure that if people try to use the upgrade scripts it fails.

Do we have revision logic in there anywhere?  That checks the subversion one
they are using 

thanks,
Regina

> -----Original Message-----
> From: Paragon Corporation [mailto:lr at pcorp.us] 
> Sent: Wednesday, December 07, 2011 3:43 PM
> To: 'PostGIS Development Discussion'
> Subject: RE: [postgis-devel] Vive Doublebox
> 
> Paul,
> 
> Okay I will not stand in your way.  In a sense I feel if I 
> played with the memory allocations things might be more even, 
> but I don't really have time for that exercise.
> I will however try to compare the speeds of my production 1.5 
> with whatever new thing you commit.
> 
> That said. You have my 
> 
> +0.33333333333333333333333333333333333333333333333333333333333
> 3333333333
> +3333333333333333333333333333333333333333333333333333333333333
> 3333333333
> +3333333333333333333333333333333333333333333333333333333333333
> 3333333333
> +33333333333333333333333333333333333333333333333
> 
> Blessing 
> 
> Thanks,
> Regina
> > -----Original Message-----
> > From: postgis-devel-bounces at postgis.refractions.net
> > [mailto:postgis-devel-bounces at postgis.refractions.net] On Behalf Of 
> > dustymugs
> > Sent: Wednesday, December 07, 2011 12:21 PM
> > To: postgis-devel at postgis.refractions.net
> > Subject: Re: [postgis-devel] Vive Doublebox
> > 
> > On 12/07/2011 08:32 AM, Paul Ramsey wrote:
> > > I was all ready to just move ahead and discard doublebox,
> > and then I
> > > realized that the next thing I'm going to move ahead on 
> (bbox type 
> > > with SRID support) will be pretty brutally impacted. One of
> > the things
> > > Sandro asked for in the bbox type was that it very quickly
> > cast from
> > > geometry. That implies reading the cached box. That implies
> > that, if
> > > we stay as we are, the result of
> > >
> > > 'LINESTRING(0 0, 1 1)'::geometry::bbox will be 'BOX(-0.00000001 
> > > -0.00000001, 1.00000001 1.000000001)'
> > >
> > > This is nothing new, it's what you get if you cast to
> > box2d, but it's
> > > more user visible cruft around having a float cache on a 
> value (the
> > > bounds) that people expect to be exactly matching the
> > coordinates. It
> > > feels a shame to make a shiny brand new type and have it visibly 
> > > reflect one of the uglier design compromises of the 1.x series.
> > >
> > > I'm tempted to apply the doublebox to trunk, and just
> > commit to trying
> > > to make it faster over time.
> > >
> > 
> > Fine by me!
> > 
> > -bborie
> > _______________________________________________
> > postgis-devel mailing list
> > postgis-devel at postgis.refractions.net
> > http://postgis.refractions.net/mailman/listinfo/postgis-devel
> > 
> 





More information about the postgis-devel mailing list