[OSM-talk] OSMXAPI

Igor Brejc igor.brejc at gmail.com
Wed Jul 23 18:57:47 BST 2008


OK I tried it with Kosmos, looks like you set the maximum size limit on 
downloadable areas (or is it just the max. count of elements)?

This is my request URL (roughly the area around Bodensee): 
http://osmxapi.hypercube.telascience.org/api/0.5/*%5bbbox=8.57056470092773,47.3201045967113,10.3134986679315,47.8968827018757

Response:
<?xml version='1.0' standalone='no'?>\n<error>\nBETA: we are testing a 
request validation mechanism to filter out silly requests.  If you have 
made\n a sensible request that is being rejected please let me know 
(80n80n at gmail.com).  \n\nYour request (*/*/*) is too large.  Please 
check your request.  \nIf you really do need this data then it may be 
better to get it directly from a planet file.\nLog ID=475746  \n</error>\n

I think it is good that you've set some sort of filtering of requests. 
It would however be nice to know what kind of criteria you set for 
rejecting requests, so that OSMXAPI clients could implements similar 
logic even before sending requests.

An idea to consider: why not allow some sort of "capabilities" query 
which would return current values of these criteria?

Regards,
Igor

80n wrote:
> I've just implemented a change to OSMXAPI which will filter out and 
> reject silly requests (like highway=*).
>
> The filter uses a combination of bbox size and count of matching 
> elements to determine whether or not to process the request.
>
> At this stage it is a bit experimental and there is scope for fine 
> tuning what it accepts and rejects.
>
> If you find that it is rejecting a request that you think is 
> reasonable then please let me know.  I want to find the point where 
> all reasonable requests are still accepted but silly or badly formed 
> ones are rejected. 
>
> This should make a big difference to the throughput as previously 
> osmxapi would easily get bogged down by someone requesting the whole 
> planet and then retrying two or three times when they don't get a 
> response within 10 seconds.
>
> 80n
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk
>   


-- 
http://igorbrejc.net





More information about the talk mailing list