[postgis-devel] "Hard" topology test case

[Laszlo Istvan]

Hi, Paul,

To continue our former discussion with a wider audience: I have 
found a small example which still causes GEOS Intersection to 
generate invalid (self-intersecting) output from valid input.

PostgreSQL 8.0.1 on i686-pc-linux-gnu

POSTGIS="1.1.2" GEOS="3.0.0cvs-CAPI-1.1.1"
PROJ="Rel. 4.4.9, 29 Oct 2004" USE_STATS

Both PostGIS and GEOS are the CVS version of 31 March.

I have included below the WKB of the two polygons (broken at 
every 80th characters). They intersect along a rather narrow 
strip.

I hope it helps in the debugging and improving the topology 
code, and I will try to find further samples of acceptable 
sizes.

Best regards,
Istvan Laszlo

On Tue, 21 Mar 2006, Paul Ramsey wrote:

> By all means, open the discussion. At this point we are really trying to get 
> as many cases as possible in our "library of really hard things".
> P
>

01030000000100000008000000AC86FF1FC6F8254100000000E89D0041380EF9FE79F925416F9092
F857A20041E6ADF7FE4BFA25417E7C8BF88FA60041A919EAFE6302264172D74AF817CD0041FB3EE8
4F640626410000000068DC004100000000DA0626410000000068DC004100000000DA062641000000
00E89D0041AC86FF1FC6F8254100000000E89D0041

01030000000100000018000000168BC5A2D7FE25410000000068C3004100000000FCFE2541000000
C0C8C20041000000E027FF25410000004065C20041000000A047FF2541000000A034C20041000000
605EFF2541000000C0EEC1004100000000BAFF2541000000A0A5C0004160E9E55DC1FF25414D5FB9
518CC0004198536655CAFF254170C769846DC00041000000209BFF2541000000A0A5BF0041000000
A076FF254100000080FEBE00410000000041FF2541000000C020BE00410000004039FF2541000000
8000BE0041000000E020FF2541000000E089BD0041F9E006050BFF25416B82F03B28BD0041000000
E0EAFE2541000000A098BC00419827B70BC3FE2541BA1E1398D1BB00410000004069FE2541000000
E010BA00410000000030FE25410000006015B9004183C61E772FFE2541271E1A0513B900411F2E34
7D17FE25414446286AA9B80041000000400FFE25410000002085B80041000000007AFD25417AAC21
E649BA0041000000007AFD25410000000068C30041168BC5A2D7FE25410000000068C30041