[Routing] monav, the very first try

[VeaaC FDIRCT]

Hi,

On Wed, Sep 1, 2010 at 4:42 PM, feverzsj <feverzsj at hotmail.com> wrote:
> After struggling with qmake and pkgconfig for a long time, I found the
> simplest way to get monav compiled under windows is directly importing the
> .pro file into VS via qt vs addin or using qmake to generate the vcproj(if
> your VS doesn't support addin). Then, add the dependencies, correct the
> resource folder(vs doesn't understand link ../../images/), change c99
> feature(dynamic array) to c90, switch on openmp language support of msvc,
> comment out qt mobile related code, and just build all.

I will try to provide Windows binaries for the preprocessor for the
next release.

> I used the preprocessor to preprocess the china.osm downloaded from
> http://downloads.cloudmade.com/asia/china#downloads_breadcrumbs. The full
> preprocess(default setting) only took several minutes(the area is not well
> mapped, so the dataset is relatively small). In the monav client, the
> AddressLookup dialog is confusing at first sight, which is mainly because of
> the area contains lots of nations and languages, and the osm importer seems
> to just import same name tag.

I am currently using the "name" tag to determine street and place
names. The current AddressLookup plugin can handle languages with a
high count of different characters, but not very well. Do you have a
suggestions how to handle this better than it is done currently? What
is the usual input method for Chinese characters? How do Chinese
satnavs solve the problem? Should I add the option to default to
English names?

> The routing is blazingly fast! Faster than most desktop/mobile routing
> engines I've ever tried, except the feature is still simple(only several
> weights:"oneway" "speed" are involved?). I set the source around "上海市 " ->
> "City Center" and target around "北京市" -> "City Center", the routing was
> immediately displayed on the map and the result seems to be what it should
> be (screenshot:
> http://fastfreefilehosting.com/file/45116/shanghai-to-beijing-jpg.html).
> Then I exchanged the source and target, but the result became some kinda
> "longest" route between two point(screenshot:
> http://FastFreeFileHosting.com/file/45117/beijing-to-shanghai-jpg.html). So,
> I repreprocessed the osm file with oneway tag ignored, and got the similar
> result as first route. That somehow confuses me. The main links between
> shanghai and beijing are motoway like
> http://www.openstreetmap.org/browse/way/65066620, and both cities should be
> well mapped.

I will look into it, but I cannot reproduce the error using the latest
planet.osm. I get the correct route in both cases. I am going to try
the osm file you used later. What version of MoNav did you use? The
0.1 release or a SVN revision? Maybe an error has crept into the
latest SVN version as I rewrote most of the compressed graph data
structure.

>
> I'd say monav is promising for fast(yet simple) routing and low ram usage.
> I'd also expect features like routing by lat/lon and reverse geocoding.

MoNav supports routing by lat/lon. Just open the map view and choose
"Goto GPS" from the menu. Then click the desired location.

What benefit would you gain from reverse Geocoding?

Greetings

Christian Vetter