[mkgmap-dev] splitter r239

Wed Nov 21 15:35:07 GMT 2012

Felix Hartmann-2 wrote
> No, The main reason being, that in long run that would probably mean to 
> prefilter all data based on per country defined criteria - meaning a lot 
> of work.
> 
> Even though for /la merdique OSM de France/ -- the french import 
> massacre, maybe in future it will be the only solution. Even though I 
> don't yet really know, what I should cut out via osmfilter...
> On 21.11.2012 07:57, GerdP wrote:

OK, today I started to collect some information regarding the algorithm that
is used
in splitter. Just to make sure that I understand the problem with france
extract:
1) a few tiles (covering parts of the atlantic) contain very few nodes (less
than 200)
2) other tiles containing Paris have too many nodes if you use
--max-nodes=1000000

The goal would be to create the tiles so that none has less than x nodes
(maybe 10% of max-nodes?)
and none has more than <max-nodes> nodes. Also, we prefer tiles with small
aspect ratio.
Correct?

Optionally, we might calculate some kind of correction factor using the
output of mkgmap.
Maybe it will be sufficient to have the file sizes of input and output files
and the areas.list
of a former splitter run.

Ciao,
Gerd

--
View this message in context: http://gis.19327.n5.nabble.com/splitter-r239-tp5736433p5736854.html
Sent from the Mkgmap Development mailing list archive at Nabble.com.