Changing in regress caused winnie to scream

[robe]

Evidentally strk decided at r11501 it was a good idea to change the regress to use built in diff. All well and good except winnie is having a few issues with that.

I'll check and see if its something in configuration or just some assumption in strk's code causing the issue:

 loader/Point .... failed (wkt testdiff expected obtained: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/test_1_diff)
 loader/PointM ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/PointZ ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/MultiPoint ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/MultiPointM ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/MultiPointZ ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/Arc ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/ArcM ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/ArcZ ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/Polygon ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/PolygonM ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/PolygonZ ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/TSTPolygon ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/TSIPolygon ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/TSTIPolygon ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/PointWithSchema ..... failed (wkb testdiff expected obtained: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/test_16_diff)
 loader/NoTransPoint .... failed (wkt testdiff expected obtained: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/test_17_diff)
 loader/NotReallyMultiPoint ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)
 loader/MultiToSinglePoint ... failed ( wkt test: running shp2pgsql output: /c/jenkins/postgis/tmp/2.1.0SVN_pg9.2_geos3.4.0dev_gdal1.10.0w64/loader.err)

[robe]

add to list of changes r11500 which is probably the main culprit

[robe]

On closer inspection this is a whitespace issue see attached.

[robe]

test_1_diff

[robe]

Okay fails on my local dev computer as well. As far as I can tell, its because the test_1_out generated by windows has CRLF breaks and the Point.select.expected has just unix line breaks.

We could change the expected files back to native mode so it matches what is generated except that wouldn't work too well when people are building from tar ball.

strk — please set your diff to ignore line break differences, and if you can PLEASE REVERT YOUR CHANGES. I can't be having these false negatives cloudind real issues when we are in beta especially when I have REAL issues to contend with.

[robe]

I meant if you can't revert and we can revisit this after we branch 2.2.

[strk]

Let's see if Winnie likes r11502

[robe]

nope. The ming system diff does support that flag though —strip-trailing-cr.

I'll try manually to run diff with that flag to see if it works.

[robe]

I just tried running this:

diff --strip-trailing-cr Point.select.expected  test_1_out > test.diff

and works fine under ming environment, but your code is still broken - its not producing the same thing

[robe]

off course by work fine I mean my .diff file is empty with that flag and not empty without that flag.

[strk]

Yep, sorry, was only checking it but not using it. r11503 should be fine