[postgis-users] Performance comparison between PostGIS's topology creation and topojson (the module)

[Sandro Santilli]

On Tue, Mar 18, 2014 at 11:04:27AM +1100, Luca Morandini wrote:
> On 18/03/14 10:28, Stephen V. Mather wrote:
> >I don't know.  Javascript is pretty fast these days. Given this and 1) no
> >database overhead, 2) Not the same level of completeness to the algorithm, I
> >would not be surprised by the JS being faster, in fact I'd expect it.
> 
> Certainly transactions are an overhead, but closeness to the data
> should be an advantage (though not that large for small datasets).

You fetch all the data into memory before turning into TopoJSON,
so the data is immediately available also to javascript.

> Actually, I presume adding one geometry at a time may be the real
> performance killer, compared with knowing in advance all the
> geometries that make the topology.

That's the other aspect of it, but as I said the PostGIS equivalent
of that (CreateTopogeo) is currently also not optimized to take
advantage of that. I'd love to see some contribution in the direction
of speeding that up :)

--strk; 

 ()  ASCII ribbon campaign  --  Keep it simple !
 /\  http://strk.keybit.net/rants/ascii_mails.txt