[osmosis-dev] bounding-polygon dismisses points at +180° longitude

[WanMil]

> On Thu, Jun 23, 2011 at 4:00 AM, WanMil <wmgcnfg at web.de
> <mailto:wmgcnfg at web.de>> wrote:
>
>     Am 15.06.2011 20:34, schrieb WanMil:
>
>         Am 14.06.2011 22:57, schrieb WanMil:
>
>                 Hi,
>
>                     Although I wrote to Brett directly that extending
>                     the polygon to 181°
>                     works I have observed that is works only for osm
>                     input but not for pbf
>                     input.
>
>                     osmosis --read-pbf file=180.osm.pbf
>                     --bounding-polygon file=asia.poly
>                     --write-xml file=filtered.osm
>                     throws an exception:
>
>                     SCHWERWIEGEND: Thread for task 1-read-pbf failed
>
>                  > [...snip...]
>                  > So the pbf parser should be changed.
>
>                 I don't think this is a problem in the PBF parser. It
>                 just shows up in
>                 the stack trace because in your pipeline, the PBF reader
>                 and the poly
>                 filter are in the same thread. The relevant portion of
>                 the stack trace
>                 is:
>
>                     java.lang.__IllegalArgumentException: Bound
>                     coordinates outside of valid
>                     range
>                     at
>                     org.openstreetmap.osmosis.__core.domain.v0_6.Bound.<init>(__Bound.java:72)
>                     at
>                     org.openstreetmap.osmosis.__areafilter.v0_6.PolygonFilter.__simpleBoundIn
>                     tersect(PolygonFilter.java:__118)
>                     at
>                     org.openstreetmap.osmosis.__areafilter.v0_6.PolygonFilter.__process(Polyg
>                     onFilter.java:71)
>
>
>                 The Bound class does not or should not really care at
>                 this point whether
>                 the data is from PBF or XML or wherever. So if there's a
>                 bug, it's
>                 actually in the Bound which has some incorrect argument
>                 validation code
>                 or in the PolygonFilter which passes incorrect arguments
>                 to the Bound,
>                 not in the PBF reader.
>
>                 Could you post your extended polygon with 181° here so I
>                 can test this?
>                 Because if it works with XML but not with PBF it gets really
>                 interesting ;)
>
>                 Bye
>                 Igor
>
>
>             Hi Igor,
>
>             attached you find the modified poly file.
>             I have to add that I tried the XML reader with my test file
>             whereas the
>             PBF reader was tested with the pbf planet dumpD
>
>             (ftp://ftp5.gwdg.de/pub/misc/__openstreetmap/planet.__openstreetmap.org/pbf-__experimental/planet-latest.__osm.pbf
>             <ftp://ftp5.gwdg.de/pub/misc/openstreetmap/planet.openstreetmap.org/pbf-experimental/planet-latest.osm.pbf>
>
>             which I think is dated from June 10th). Maybe that makes a
>             difference?
>
>             PBF reading failed with the given exception. XML reading was
>             successful
>             and the 180° longitude node was successfully transferred to
>             the filtered
>             file.
>
>             WanMil
>
>
>         I have found the reason for the problem. It's not pbf related.
>
>         Attached osm file also throws the exception. The reason is the
>         bounding
>         box information which is not accepted:
>         <bound box="-90.00000,-180.00000,90.__00000,180.00000"
>         origin="http://www.__openstreetmap.org/api/0.6
>         <http://www.openstreetmap.org/api/0.6>"/>
>
>         WanMil
>
>
>     Are there any plans to fix that problem? From my point of view it
>     should be easily fixable by removing the check in the Bound class or
>     if the check is less strict. But I don't know if there are any side
>     effects.
>
>
> Perhaps we could simply relax the Bound checks to allow slightly larger
> values such as 181 and 91.  It's ugly, but I don't have a better idea.
> At least it should catch the majority of error cases.
>
> Any objections?
>
> Brett
>

Hi,
what need to be done to get this issue fixed?
The following things were provided:
* a simple testcase
* a scenario in which the problem is a real issue (extracting russian 
border)
* a simple patch based on Brett ideas

I don't know what I can do now to speed up a bugfix. So please commit 
the patch or let me know what's wrong with it.

Thanks
WanMil