[postgis-users] Changed bahaviour of PostGIS confuses MapServer

[Paul Ramsey]

Yes, Peter, I think it should call for a quick 1.3.5. Mapserver users
are a huge contingent of PostGIS users, and anyone with a LINE layer
now has a crasher. 1.3.5 should come off the line ASAP. There's always
more version numbers, IMO.

Can we have a quick vote on 1.3.5 release-from-branch, no-waities?

+1

P.

On Sat, Dec 13, 2008 at 1:25 AM, Peter Hopfgartner
<peter.hopfgartner at r3-gis.com> wrote:
> Mark Cave-Ayland wrote:
>>
>> Peter Hopfgartner wrote:
>>
>>> Dear MLs,
>>>
>>> sorry for cross-posting to both, PostGIS and MS, but seems to rtouch
>>> both.
>>>
>>> I've tried to install PostGIS 1.3.4 on a CentOS 5 server with MapServer
>>> 5.2.1 and PostgreSQL 8.2.11.
>>>
>>> the command:
>>>
>>> shp2img -l prova -e 1587541 5116731 1591730 5120008 -m prova.map >
>>> ../../public/technet/test-<pgis-version>.png
>>>
>>> with 1.3.3 gave the result shown in figure test-postgis-1.3.3-4.png (a
>>> simple line layer)
>>>
>>> Upgrading to 1.3.4 the picture became as in test-postgis-1.3.4-1.png, an
>>> empty picture.
>>>
>>> The relevant mapfile fragment is Mapfile.frag.txt, the SQL table
>>> definition is e405.sql, all objects are MULTILINESTRINGs [SELECT DISTINCT
>>> geometrytype(the_geom) FROM public.e405 results in MULTILINESTRING], the
>>> geometry column is registered as MULTILINESTRING in geometry_columns.
>>>
>>> Running the MapServer query:
>>>
>>> SELECT
>>> asbinary(force_collection(force_2d(the_geom)),'NDR'),common_uid::text from
>>> public.e405 WHERE (tiplinapp is not null) and (the_geom && setSRID(
>>> 'BOX3D(1587541 5116702.44897959,1591730
>>> 5120036.55102041)'::BOX3D,find_srid('','public.e405','the_geom') ))
>>>
>>> changed for better result readability into:
>>>
>>> SELECT AsEWKT(force_collection(force_2d(the_geom))),common_uid::text from
>>> public.e405 WHERE (tiplinapp is not null) and (the_geom && setSRID(
>>> 'BOX3D(1587541 5116702.44897959,1591730
>>> 5120036.55102041)'::BOX3D,find_srid('','public.e405','the_geom') ))
>>>
>>> by hand gives slightly different results, as shown in the attached files.
>>>
>>> To make a long story short, 1.3.4 seems to better preserve the
>>> MULTILINESTRING, but this seems to confuse MapServer.
>>>
>>> Regards,
>>>
>>> Peter
>>
>>
>> Hi Peter,
>>
>> This is likely the known issue with MULTILINESTRINGs in 1.3.4 causing
>> segfaults :(  The patch you need is very simple and can be found here:
>> http://postgis.refractions.net/pipermail/postgis-commits/2008-December/000866.html
>> if you can't wait for a 1.3.5 release. Please let us know whether this
>> resolves the problem for you.
>>
>>
>> HTH,
>>
>> Mark.
>>
> Should this issue call for a quick 1.3.5 release? If I understood this issue
> well, everybody rendering PostGIS data stored as a MULTILINE geometry will
> have this very same problem, at least with MapServer. I would guess that
> this is quite a big share of PostGIS users that will be hit by this issue.
>
> Regards,
>
> Peter
>
>
>
>
> --
>
> Dott. Peter Hopfgartner
>
> R3 GIS Srl - GmbH
> Via Johann Kravogl-Str. 2
> I-39012 Meran/Merano (BZ)
> Email: peter.hopfgartner at r3-gis.com
> Tel. : +39 0473 494949
> Fax  : +39 0473 069902
> www  : http://www.r3-gis.com
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-users
>