cunit failure on mingw64 rc1 - 1.#INF

[robe]

I think there was a time when all my cunit tests used to pass, but it's possible it was failing on beta4 too since I think I had to cd to the other folders to tests the other tests. So this might be cause of a change in my mingw64 environment. I think I was running gcc 4.4.7 or something when I was paying more attention and everything was passing.

Suite: print_suite
  Test: test_lwprint_default_format ...passed
  Test: test_lwprint_format_orders ...passed
  Test: test_lwprint_optional_format ...passed
  Test: test_lwprint_oddball_formats ...passed
  Test: test_lwprint_bad_formats ...passed
Suite: misc
  Test: test_misc_force_2d ...passed
  Test: test_misc_simplify ...passed
  Test: test_misc_count_vertices ...passed
  Test: test_misc_area ...passed
  Test: test_misc_wkb ...passed
Suite: ptarray
  Test: test_ptarray_append_point ...passed
  Test: test_ptarray_append_ptarray ...passed
  Test: test_ptarray_locate_point ...passed
  Test: test_ptarray_isccw ...passed
  Test: test_ptarray_desegmentize ...passed
  Test: test_ptarray_insert_point ...passed
Suite: PostGIS Computational Geometry Suite
  Test: test_lw_segment_side ...passed
  Test: test_lw_segment_intersects ...passed
  Test: test_lwline_crossing_short_lines ...passed
  Test: test_lwline_crossing_long_lines ...passed
  Test: test_lwline_crossing_bugs ...passed
  Test: test_lwpoint_set_ordinate ...passed
  Test: test_lwpoint_get_ordinate ...passed
  Test: test_point_interpolate ...passed
  Test: test_lwline_clip ...passed
  Test: test_lwline_clip_big ...passed
  Test: test_lwmline_clip ...passed
  Test: test_geohash_point ...passed
  Test: test_geohash_precision ...passed
  Test: test_geohash ...passed
  Test: test_isclosed ...passed
Suite: PostGIS Measures Suite
  Test: test_mindistance2d_tolerance ...passed
  Test: test_rect_tree_contains_point ...passed
  Test: test_rect_tree_intersects_tree ...passed
  Test: test_lwgeom_segmentize2d ...passed
Suite: node
  Test: test_lwgeom_node ...passed
Suite: WKT Out Suite
  Test: test_wkt_out_point ...passed
  Test: test_wkt_out_linestring ...passed
  Test: test_wkt_out_polygon ...passed
  Test: test_wkt_out_multipoint ...passed
  Test: test_wkt_out_multilinestring ...passed
  Test: test_wkt_out_multipolygon ...passed
  Test: test_wkt_out_collection ...passed
  Test: test_wkt_out_circularstring ...passed
  Test: test_wkt_out_compoundcurve ...passed
  Test: test_wkt_out_curvpolygon ...passed
  Test: test_wkt_out_multicurve ...passed
  Test: test_wkt_out_multisurface ...passed
Suite: WKT In Suite
  Test: test_wkt_in_point ...FAILED
    1. cu_in_wkt.c:71  - CU_ASSERT_STRING_EQUAL(r,"POINT(inf 0)")
  Test: test_wkt_in_linestring ...passed
  Test: test_wkt_in_polygon ...passed
  Test: test_wkt_in_multipoint ...passed
  Test: test_wkt_in_multilinestring ...passed
  Test: test_wkt_in_multipolygon ...passed
  Test: test_wkt_in_collection ...passed
  Test: test_wkt_in_circularstring ...passed
  Test: test_wkt_in_compoundcurve ...passed
  Test: test_wkt_in_curvpolygon ...passed
  Test: test_wkt_in_multicurve ...passed
  Test: test_wkt_in_multisurface ...passed
  Test: test_wkt_in_tin ...passed
  Test: test_wkt_in_polyhedralsurface ...passed
  Test: test_wkt_in_errlocation ...passed
Suite: WKB Out Suite
  Test: test_wkb_out_point ...passed
  Test: test_wkb_out_linestring ...passed
  Test: test_wkb_out_polygon ...passed
  Test: test_wkb_out_multipoint ...passed
  Test: test_wkb_out_multilinestring ...passed
  Test: test_wkb_out_multipolygon ...passed
  Test: test_wkb_out_collection ...passed
  Test: test_wkb_out_circularstring ...passed
  Test: test_wkb_out_compoundcurve ...passed
  Test: test_wkb_out_curvpolygon ...passed
  Test: test_wkb_out_multicurve ...passed
  Test: test_wkb_out_multisurface ...passed
  Test: test_wkb_out_polyhedralsurface ...passed
Suite: WKB In Suite
  Test: test_wkb_in_point ...passed
  Test: test_wkb_in_linestring ...passed
  Test: test_wkb_in_polygon ...passed
  Test: test_wkb_in_multipoint ...passed
  Test: test_wkb_in_multilinestring ...passed
  Test: test_wkb_in_multipolygon ...passed
  Test: test_wkb_in_collection ...passed
  Test: test_wkb_in_circularstring ...passed
  Test: test_wkb_in_compoundcurve ...passed
  Test: test_wkb_in_curvpolygon ...passed
  Test: test_wkb_in_multicurve ...passed
  Test: test_wkb_in_multisurface ...passed
  Test: test_wkb_in_malformed ...passed
Suite: libgeom
  Test: test_typmod_macros ...passed
  Test: test_flags_macros ...passed
  Test: test_serialized_srid ...SRID value -3005 converted to the offici
nown SRID value 0
passed
  Test: test_gserialized_from_lwgeom_size ...passed
  Test: test_gbox_serialized_size ...passed
  Test: test_lwgeom_from_gserialized ...passed
  Test: test_lwgeom_count_vertices ...passed
  Test: test_on_gser_lwgeom_count_vertices ...passed
  Test: test_geometry_type_from_string ...passed
  Test: test_lwcollection_extract ...passed
  Test: test_lwgeom_free ...passed
  Test: test_lwgeom_flip_coordinates ...passed
  Test: test_f2d ...passed
  Test: test_lwgeom_clone ...passed
  Test: test_lwgeom_force_clockwise ...passed
  Test: test_lwgeom_calculate_gbox ...passed
  Test: test_lwgeom_is_empty ...passed
  Test: test_lwgeom_same ...passed
Suite: split
  Test: test_lwline_split_by_point_to ...passed
  Test: test_lwgeom_split ...passed
Suite: Geodetic Suite
  Test: test_signum ...passed
  Test: test_gbox_from_spherical_coordinates ...passed
  Test: test_gserialized_get_gbox_geocentric ...passed
  Test: test_clairaut ...passed
  Test: test_edge_intersection ...passed
  Test: test_edge_distance_to_point ...passed
  Test: test_edge_distance_to_edge ...passed
  Test: test_lwgeom_distance_sphere ...passed
  Test: test_lwgeom_check_geodetic ...passed
  Test: test_gserialized_from_lwgeom ...passed
  Test: test_spheroid_distance ...passed
  Test: test_spheroid_area ...passed
  Test: test_lwpoly_covers_point2d ...passed
  Test: test_ptarray_point_in_ring ...passed
Suite: GEOS
  Test: test_geos_noop ...passed
Suite: stringbuffer
  Test: test_stringbuffer_append ...passed
  Test: test_stringbuffer_aprintf ...passed
Suite: surface
  Test: triangle_parse ...passed
  Test: tin_parse ...passed
  Test: polyhedralsurface_parse ...passed
  Test: tin_tgeom ...passed
  Test: psurface_tgeom ...passed
  Test: surface_dimension ...passed
  Test: surface_perimeter ...passed
Suite: Homogenize Suite
  Test: test_coll_point ...passed
  Test: test_coll_line ...passed
  Test: test_coll_poly ...passed
  Test: test_coll_coll ...passed
  Test: test_geom ...passed
  Test: test_coll_curve ...passed
Suite: out_gml
  Test: out_gml_test_precision ...passed
  Test: out_gml_test_srid ...passed
  Test: out_gml_test_dims ...passed
  Test: out_gml_test_geodetic ...passed
  Test: out_gml_test_geoms ...passed
  Test: out_gml_test_geoms_prefix ...passed
  Test: out_gml_test_geoms_nodims ...passed
  Test: out_gml2_extent ...passed
  Test: out_gml3_extent ...passed
Suite: KML Out Suite
  Test: out_kml_test_precision ...passed
  Test: out_kml_test_dims ...passed
  Test: out_kml_test_geoms ...passed
  Test: out_kml_test_prefix ...passed
Suite: GeoJson Out Suite
  Test: out_geojson_test_precision ...passed
  Test: out_geojson_test_dims ...passed
  Test: out_geojson_test_srid ...passed
  Test: out_geojson_test_bbox ...passed
  Test: out_geojson_test_geoms ...passed
Suite: SVG Out Suite
  Test: out_svg_test_precision ...passed
  Test: out_svg_test_dims ...passed
  Test: out_svg_test_relative ...passed
  Test: out_svg_test_geoms ...passed
  Test: out_svg_test_srid ...passed
Suite: X3D Out Suite
  Test: out_x3d3_test_precision ...passed
  Test: out_x3d3_test_geoms ...passed

running gcc.exe (GCC) 4.5.4 20111030 (prerelease)

no reason to stop the show for this one I don't think. I'll upgrade and see if I still have same issue.

If I put in a printf to see the difference I get:

IN:  POINT(1e700 0)
OUT: POINT(1.#INF 0)

so I suppose it's probably a harmless failure

[robe]

I'm sure I'm the only one who suffers from this misfortune. Is there any way to get POINT(inf 0) and POINT(1.#INF 0) to be treated the same.

Every time I compile for windows 64-bit I have to remark out this test so regression checks continue.

[mwtoews]

This is compiler-specific, defined in a macro INFINITY in math.h for C99 compilers.

However, there is a good hint on line 88 of regress.sql, which could be modified to use a better regexp:

select '62', regexp_replace(ST_asewkt('POINT( 1e700 0)'::GEOMETRY), '(Infinity|1\.#INF)', 'inf');

[strk]

@robe : do you also get the failure with this:

cd regress && ./run_test -v regress 

[strk]

Please try r9613 (and see r9612 too) — reopen if not fixed yet

[robe]

I wonder if we have some extra regex in the normal postgresql checks (like we do for the cleanup I know we had some for some other idiosyncracies that strips the test file after the fact). Surprisingly the regress.sql has never been failing so actually no need to change that.

However if I run :

select '63',ST_asewkt('POINT( -1e700 0)'::GEOMETRY);

I get:

63  POINT(-1.#INF 0)

Have no clue how that is passing regress. anyrate I haven't upgraded yet, but the 9612 fix doesn't appear to be necessary.

[strk]

Evidently postgresql shades Inf reporting away. The commit won't hurt anyway…