[raster] regress failure on mingw mapalgebraexpr, rt_clip

[robe]

Not sure how long this has been an issue. I got halted checking because of the build area regresses and was too lazy to move on.

Anyrate on my pg 8.4 mingw32bit gcc 3.4.5, I am getting regression failures.

 rt_mapalgebraexpr .. failed (diff expected obtained: /tmp/pgis_reg/test_41_diff)
 rt_mapalgebrafct .. ok 
 rt_mapalgebraexpr_2raster .. ok 
 rt_mapalgebrafct_2raster .. ok 
 drop_rt_mapalgebra_test .. ok 
 create_rt_mapalgebrafctngb_test .. ok 
 rt_mapalgebrafctngb .. ok 
 rt_mapalgebrafctngb_userfunc .. ok 
 drop_rt_mapalgebrafctngb_test .. ok 
 rt_reclass .. ok 
 rt_resample .. ok 
 rt_asraster .. ok 
 rt_intersection .. ok 
 rt_clip .. failed (diff expected obtained: /tmp/pgis_reg/test_54_diff)
 create_rt_gist_test .. ok 
 rt_above .. ok 
 rt_below .. ok 
 rt_contained .. ok 
 rt_contain .. ok 
 rt_left .. ok 
 rt_overabove .. ok 
 rt_overbelow .. ok 
 rt_overlap .. ok 
 rt_overleft .. ok 
 rt_overright .. ok 
 rt_right .. ok 
 rt_same .. ok 
 drop_rt_gist_test .. ok 
 rt_spatial_relationship .. ok 
 rt_intersects .. ok 
 rt_samealignment .. ok 
 bug_test_car5 .. ok 
 tickets .. ok 
 loader/Basic ...... ok 
 loader/BasicCopy ...... ok 
 loader/Tiled10x10 ..... ok 
 loader/Tiled10x10Copy ..... ok 
 uninstall ... ok (3699)

Run tests: 78
Failed: 2

using gdal 1.9.0. I have to test the others to make sure its not just something screwy with my envs.

See attached files

[dustymugs]

I'll take a look as soon as I can (after June 1… sorry).

[dustymugs]

Strange set of regression failures as the functions failing aren't related and functions related to the ones failing aren't.

The mapalgebraexpr failure appears to be caused by a rounding issue. The specific test that is failing is rounding a pixel value of 27.5, which could go to 27 or 28 depending on the platform. I'm going to change the round() to a ceil() as that should be more consistent across all platforms.

It looks like the output queries in rt_clip aren't explicitly ordered like they should.

Can you try r9858 for 2.0? I've also committed the change for trunk in r9858.

[dustymugs]

Punting to robe for testing

[robe]

Back to you dustymugs.

It's better. rt_clip is still failing.

rt_clip .. failed (diff expected obtained: /tmp/pgis_reg/test_54_diff)

I've attached the diff.

[dustymugs]

Still looks like an ORDER BY issue to me. Can you try r9912? Trunk also changed in r9913

[robe]

Sadly still fails on r9913. As I recall, works fine on my 9.0 and 9.1 so something about 8.4 (not just a windows issue though that may be part). I'll double-check the others. You don't have a 8.4 per chance to tst on to rule out something funk with my 8.4 config.

[robe]

Hmm I think strk still has 8.4. strk having any regression issues with raster on 8.4 or are you not regress testing raster?

[dustymugs]

Testing on 8.4 and I see the same issue. So not just mingw.

[dustymugs]

Alright. Regression tests pass in PG 9.1, 9.0 and 8.4. Committed in r9914 for 2.0 and r9915 for 2.1.

[robe]

Great fixes my issue too.