[OSM-talk] OSM maps on Garmin GPS

[SteveC]

* @ 21/11/06 04:31:03 PM steve at asklater.com wrote:
> * @ 21/11/06 04:31:25 PM sxpert at sxpert.org wrote:
> > SteveC wrote:
> > >* @ 21/11/06 04:18:30 PM sxpert at sxpert.org wrote:
> > >>well, lets put it this way, necessity is the mother of invention.
> > >>if people came up with this, there must be a good reason, such as 
> > >>"yikes, it's faster".
> > >>what would have been the reason if it didn't give any advantage over the 
> > >>regular thing ?
> > >
> > >So you're asserting it's better because it must me?
> > >
> > >Like, everyone uses Windows so it must be better?
> > 
> > I never asserted this.
> > I said that there must be a reason that they came up with that, what 
> > would be the point of researching this otherwise.
> 
> Because the SQL looks prettier?
> 
> > now, if you can prove that using 2 columns and 2 normal btree indexes is 
> > faster than using an r-tree index on the set of 2 (or more) columns, I'm 
> > listening :D
> 
> Like I said, I thought Nick Hill did. But you are the spatial bandwagon

Specificially

http://lists.openstreetmap.org/pipermail/dev/2006-September/002077.html

and other posts/threads on the subject of lat/lon storage and indexing.

I'm _totally_ up for using spatial columns or anything else if it helps,
but not just because thats what everyone else uses. If spatial is faster
lets use it, but a proof _is_ required.

have fun,

SteveC steve at asklater.com http://www.asklater.com/steve/