[postgis-devel] [postgis-users] Infinite loop in st_intersects - because of incorrect data out of st_transform?

[Sandro Santilli]

On Thu, May 12, 2011 at 09:38:04AM -0700, Paul Ramsey wrote:
> On Thu, May 12, 2011 at 1:35 AM, Sandro Santilli <strk at keybit.net> wrote:
> =
> > Not sure we want to refuse to accept inf coordinates though,
> > as long as ST_IsValid can catch them, but most importantly
> > as long as you can have them in your backups...
> 
> What's the use case for Inf coordinates? Or NaN coordinates? There's
> no real-world ones, are there? and once we're painted into that corner
> computationally I'm not sure how we get back out.

It is indeed hard to figure how to get out of there, but my point
is refusing to output such things would prevent us from backing up
the data and refusing them in input would prevent us from restoring
backups. Once they are in, somehow, you can't stop accepting them
and outputting them.

--strk; 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html