[raster] Raster some check fails

[aperi2007]

Hi, just executing "make raster-post-install-check" I see these errors:

./run_test ...
Creating spatial db rtpostgis_reg
createlang: linguaggio "plpgsql" già installato nel database "rtpostgis_reg"
TMPDIR is /tmp/rtpgis_reg_11269

 PostgreSQL 9.0.2 on i686-pc-linux-gnu, compiled by GCC gcc (Debian 4.3.2-1.1) 4.3.2, 32-bit
 Postgis 2.0.0SVN - 2011-01-10 18:26:09
   GEOS: 3.3.0-CAPI-1.7.0
   PROJ: Rel. 4.7.1, 23 September 2009
 WKTRaster 2.0.0SVN - 2011-01-10 18:26:09

Running tests

 check_raster_columns. ok
 check_raster_overviews. ok
 rt_io. ok
 create_rt_bytea_test. ok
 rt_bytea. ok
 create_rt_box2d_test. ok
 rt_box2d. ok
 rt_addband. failed (diff expected obtained: /tmp/rtpgis_reg_11269/test_8_diff)
 create_rt_properties_test. ok
 rt_dimensions. ok
 rt_scale. ok
 rt_upperleft. ok
 rt_rotation. ok
 rt_georeference. ok
 rt_set_properties. ok
 create_rt_empty_raster_test. ok
 rt_isempty. ok
 rt_hasnoband. ok
 create_rt_band_properties_test. ok
 rt_band_properties. failed (diff expected obtained: /tmp/rtpgis_reg_11269/test_20_diff)
 rt_set_band_properties. ok
 rt_pixelvalue. ok
 create_rt_utility_test. ok
 rt_utility. ok
 create_rt_gist_test. failed (diff expected obtained: /tmp/rtpgis_reg_11269/test_25_diff)
 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
 rt_spatial_relationship. ok
 bug_test_car5. ok

Run tests: 39
Failed: 3
make[3]: Leaving directory `/home/postgres/software/postgis/postgis-2.0.0SVN/raster/test/regress'
make[2]: Leaving directory `/home/postgres/software/postgis/postgis-2.0.0SVN/raster/test'
make[1]: Leaving directory `/home/postgres/software/postgis/postgis-2.0.0SVN/raster'
postgres@deb:~/software/postgis/postgis-2.0.0SVN$
-----

I add the tmp logs.

[aperi2007]

Diff logs

[pracine]

The two first diff were just fixed. I don't get the third diff. Could you give it a try again?

[aperi2007]

Hi,

I tested. But unfortunately the error is still here.

executing: make raster-post-install-check.

...

create_rt_utility_test. ok rt_utility. ok create_rt_gist_test. failed (diff expected obtained: /tmp/rtpgis_reg_11895/test_25_diff) rt_above. ok rt_below. ok

....

I add a zip with the new diff log.

[aperi2007]

a new zip with logs

[mloskot]

Description format

[mloskot]

Some of the failures like rt_addband is related to the problem #792

I tried to run the raster-post-install-check for r6848 and getting even more failures:

PostgreSQL 8.4.7 on x86_64-pc-linux-gnu, compiled by GCC gcc-4.4.real (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5, 64-bit
 Postgis 2.0.0SVN - 2011-02-19 01:10:53
   GEOS: 3.3.0-CAPI-1.7.0
   PROJ: Rel. 4.7.1, 23 September 2009
 WKTRaster 2.0.0SVN - 2011-02-19 01:10:53

Running tests

 check_raster_columns. ok
 check_raster_overviews. ok
 rt_io. ok
 create_rt_bytea_test. ok
 rt_bytea. ok
 create_rt_box2d_test. ok
 rt_box2d. ok
 rt_addband. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_8_diff)
 create_rt_properties_test. ok
 rt_dimensions. ok
 rt_scale. ok
 rt_upperleft. ok
 rt_rotation. ok
 rt_georeference. ok
 rt_set_properties. ok
 create_rt_empty_raster_test. ok
 rt_isempty. ok
 rt_hasnoband. ok
 create_rt_band_properties_test. ok
 rt_band_properties. ok
 rt_set_band_properties. ok
 rt_pixelvalue. ok
 create_rt_utility_test. ok
 rt_utility. ok
 create_rt_mapalgebra_test. ok
 rt_mapalgebra. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_26_diff)
 create_rt_gist_test. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_27_diff)
 rt_above. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_28_diff)
 rt_below. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_29_diff)
 rt_contained. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_30_diff)
 rt_contain. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_31_diff)
 rt_left. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_32_diff)
 rt_overabove. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_33_diff)
 rt_overbelow. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_34_diff)
 rt_overlap. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_35_diff)
 rt_overleft. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_36_diff)
 rt_overright. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_37_diff)
 rt_right. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_38_diff)
 rt_same. failed (diff expected obtained: /tmp/rtpgis_reg_20787/test_39_diff)
 rt_spatial_relationship. ok
 bug_test_car5. ok

Run tests: 41
Failed: 15

AFAIU, most of them seem to be related to SQL functions renaming.

[strk]

I'm gettin failures too, with r6888 I guess the lack of a pre-install check is making these tests kind of ineffective.

 rt_addband. failed (diff expected obtained: /tmp/rtpgis_reg_24868/test_8_diff)
 rt_overlap. failed (diff expected obtained: /tmp/rtpgis_reg_24868/test_35_diff)
 rt_spatial_relationship. failed (diff expected obtained: /tmp/rtpgis_reg_24868/test_40_diff)

Run tests: 41
Failed: 17

about to attach rtpgis_reg_24868

[jorgearevalo]

There's some debug information in you output. Did you configure your sources with '--enable-debug' option?

[strk]

Ah, it's my ~/.psqlrc setting min_client_messages to DEBUG. The test script should make sure min_client_messages is set to the default instead, or that the users's .pgsqlrc isn't used.

Porting the whole infrastructure onto the run_test of core should include such facilities, I've extended it recently to allow testing topology, doing the same for raster shouldn't be too much work

[pracine]

Some of these failures are fixed at the same time as #792. The rest should be fixed by desactivating DEBUG in run_test as strk suggested. I reassign to Jorge.

[jorgearevalo]

Raster test infrastructure ported to general test infrastructure, in the same way topology did it. Anyway, I have a doubt I'd like to solve before closing the ticket:

Core run_test forces --no-psqlrc. So, if my .psqlrc sets min_client_messages to debug, run_test will avoid it. But executing configure --enable-debug makes tests fail because too much debug information sent from server. Not only raster tests.

So, I think there's a confussion here, or I'm wrong. Correct me if so:

• On one hand, run_test avoid user defined log level. This is, at client level (modifying psql call parameters).
  
• On the other hand, you can specify "--enable-debug" when you compile PostGIS. This will enable debug messages at server level (log messages sent by postgres backend when executing PostGIS tests). If you specify this option, tests will fail, because run_test only controls client debug level, not server debug level (obviuosly, because it's using psql client).
  

So, this shouldn't be a defect. It's normal. If I'm not wrong...

May I close the ticket?

[strk]

The problem with --enable-debug is that liblwgeom only exposes two message hooks: error and notice. So debugging is implemented in terms of NOTICE, rather than DEBUG. It's not related to this bug, may be worth discussing in a separate thread (mailing list or new ticket).

I think this ticket can be closed. I still have a failure in mapalgebra but I think it's worth a separate ticket.

[jorgearevalo]

Ok. I close this ticket. The problem with MapAlgebra is the one you've posted in ticket #896, I guess.