[OSM-talk] OSMXAPI question: is it me, or...

[Lambertus]

The bbox of a regional cycle relation covers only a small area e.g. 10km 
square in most situations (50km tops) and checking that against the 
requested bbox would indeed give some routes outside the requested bbox 
but certainly not from a few 100 of km away. So I don't think the 
observed behaviour is because of the rcn bbox matching the requested bbox...

80n wrote:
> On Thu, Sep 11, 2008 at 8:27 AM, Lambertus <osm at na1400.info> wrote:
>   
>> Thanks for the quick action. The amount of data downloaded by the query was
>> significantly more then before so that was already promising. Looking at the
>> data in JOSM showed no sign of crippled roads so that bug appears to be
>> squashed.
>>
>> However the bbox parameter still does not seem to do what I expect from it
>> (data returning from UK, NL, DE and BE instead of roughly a province in NL).
>>     
>
> Given that relations can contain an arbitrary collection of
> unconnected ways the notion of where a relation is might be a bit
> tricky.
>
> The current algorithm draws a box around the extent of each relation
> and then tests whether that overlaps with the bbox you have requested.
>
> Having thought about it for a moment it should probably test the
> extent of each way within the relation.  This would give better
> precision.
>
> I'm not sure if this is the cause of what you are seeing, but it seems likely.
>
>
>   
>> Reversing the parameters gives almost but not exactly the same result:
>>
>> -rw-rw-r-- 1 lambertus lambertus 16519465 Sep 11 08:45 data.osm.4 (wget
>> http://www.informationfreeway.org/api/0.5/relation[network=rcn][bbox=5.5,52.0,6.3,52.3])
>> -rw-rw-r-- 1 lambertus lambertus 16397318 Sep 11 09:14 data.osm.5 (wget
>> http://www.informationfreeway.org/api/0.5/relation[bbox=5.5,52.0,6.3,52.3][network=rcn])
>>     
>
> The order of the predicates should not make any difference at all.
> All I can think is that the database had changed during that time, so
> you would have got the same results either way.
>
>   
>> The one that was downloaded half an hour later is a bit smaller to my
>> surprise. I tried to find out what was the difference but visually there are
>> no changes to spot, but diff outputs so much that I don't know what to do
>> with it.
>>
>> 80n wrote:
>>     
>>> I've found a bug and fixed it.  Can you try again and let me know how it
>>> goes?
>>>
>>> 80n
>>>
>>>
>>>