[Routing] monav, RAM and disk usage

[Tim Teulings]

Hello!

> I doubt, that the ContractionHierarchies routing plugin of MoNav could
> save much code by using libosmscout's infrastructure. As the
> algorithm's speed heavily depends on its preprocessed data and
> specialized external memory compressed graph data structure it would
> still need to manage its own set of files.

Which is completely OK. There is still a lot of stuff you have to add  
to a routing core to make a completing routing application for this.

What comes to my mind:
* Index object by name
* Map drawing including styles etc...
* Platform independent file format
* Getting fine written textual driving orders from a list of nodes.
* Ornaization of map files

OpenStreetMap has a number of nice implementations for various  
components, but it is IMHO still lacking a framework to glue it all  
together - especially for offline, mobile backends. If people do not  
team up we likely do not get past being 60%, 70% perfect.

> Furthermore it seems that libosmscout uses Cairo for rendering. This
> would limit the supported platforms, e.g., no Windows Mobile port of
> Cairo is known to me.

It also supports Qt (which is however much slower than Qt) and is  
designed to implement other backends fast (see my other mail).

> Among other things MoNav currently offers several features that no
> other routing solution does. Therefore I want to maintain and develop
> it as a fully fledged routing solution.

Your code, your choice :-) As a demonstration code is good to have a  
as small as possile code base. But is it your plan to write a complete  
navigation solution to see your algorithm work in practice on your  
moms navigation device on your own. When would you be finished?

> If you want to integrate the ContractionHierarchies routing plugin
> into libosmscout you are welcome of course. If you run into any kind
> of problems or need help understanding the algorithm I would be happy
> to assists you.

I possibly do this. But since my time is limited I have to do one  
thing after another and currently I try to get map drawing in a  
"good-enough" state to be usable.

If 5 experts team up to put their better than average solution (and  
possibly better than commercial alternatives) into one framework,  
there should be enormous functionality after one year. If there work  
on it each on their own for the same year, output will likely not that  
impressive for other people.

-- 
Gruß...
    Tim