[GraphHopper] Why not using OSM way uid?

[Andreas Fischer]

>> non deterministic edge IDs.

> BTW: these IDs are deterministic in a sense that an identical OSM file
produces exactly the same graph file with the same IDs. No random behaviour
involved!

Yes, by non deterministic i meant, that if the OSM data changes i have no
guarantee that the ids will be the same! ;)
But that answered my question, thank you.


2014-02-17 10:49 GMT+01:00 Peter K <peathal at yahoo.de>:

>  Exactly, an edge is just a connection between two junctions.
>
> Also have a look into the Java Magazine article<http://karussell.wordpress.com/2014/01/23/graphhopper-news-article-in-java-magazine-and-fosdem-2014/>where I described this issue. E.g. using osm ids would require us to use a
> HashMap which requires more RAM (in an array you don't need to store the
> indices!) and is potentially slower than an array.
>
> If you need them nevertheless you can create a file on disc while import
> to map internal ints to OSM long IDs
>
> > non deterministic edge IDs.
>
> BTW: these IDs are deterministic in a sense that an identical OSM file
> produces exactly the same graph file with the same IDs. No random behaviour
> involved!
>
> Regards,
> Peter.
>
>   I was just wondering why the OSM UIDs are not used instead of iterative
> self generated non deterministic edge IDs.
>
>  Because a single OSM way ID might yield more than one edge?
>
> Bye
> Frederik
>
>
>
>
> _______________________________________________
> GraphHopper mailing list
> GraphHopper at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/graphhopper
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/graphhopper/attachments/20140217/c871cf66/attachment.html>