logo separator

[mkgmap-dev] Mkgmap dropping data: Area too small to split at...

From WanMil wmgcnfg at web.de on Sat Jun 18 19:07:00 BST 2011

> Hi!
>
>
> WanMil wrote:
>>
>> The only possible solution is to split the polygon so that it contains
>> less points (either by selecting the splitter split points or by
>> changing the osm data). But that's a try and error solution.
>>
>
> Thanks for the information. In that thread you have been discussion
> subdivision a lot, but to me it looks like mkgmap does not do any
> subdivision on the objects themselves - it just keeps making the area
> smaller but adding all items to the first matching area. As the objects are
> very large, this does not help, they all keep being assinged to the first
> area until that becomes too small.
>
> So I take it the patch you created in January was never included?

Yes, it's not included.

>
> Have you any more experiences with your patch? You mentioned that it might
> not work? Are you still using it? Do you think it might work with the
> current version?

No, I am not using it. Things are more complicated so the patch was a 
first try only - not more.

>
> Changing the osm data is no way to go - on the opposite, people keep
> creating more and more such huge multipolygons. Splitting them before
> calling mkgmap is what I have been trying - but I found no way to tell what
> items will cause a problem.
>
> The only practical solution seems to be that mkgmap actually splits up the
> objects instead of choking on them. Is that what your patch was doing?

I cannot remember in detail what the patch did. But I remember that I 
came to the conclusion that the whole creation of the subdivisions need 
to be reorganized. I only started with that but did not continue for a 
long time.

Maybe there is an easy patch that fixes the current problem. I'll try to 
have a look on that within the next days.
Can you upload the osm file so that it is easier for me to start?

WanMil

>
> bye
>                       Nop



More information about the mkgmap-dev mailing list