[GRASS-dev] Failing tests overview for September 2015

wenzeslaus · October 3, 2015, 4:40pm

t.rast.accumulate: created file differ from the reference

Needs investigation.

./temporal/t.rast.accumulate/testsuite/test.t.rast.accumulate.sh

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-10-03-07-00/report_for_nc_basic_spm_grass7_nc/temporal/t.rast.accumulate/test.t.rast.accumulate/index.html

t.rast.to.rast3: r3.univar ended successfully but we were expecting output and got nothing

The obvious issue here is that r3.univar (as well as r.univar) gives no output when only NULLs are present. This is strange and undocumented behavior. If the NULLs on the input are OK in this context, I don’t know.

./temporal/t.rast.to.rast3/testsuite/test_strds_to_rast3.py

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-10-03-07-00/report_for_nc_basic_spm_grass7_nc/temporal/t.rast.to.rast3/test_strds_to_rast3/index.html

pygrass/modules: Execution of subprocesses was not successful

Some problem with grid module which works otherwise, possibly just a problem with test. However, worth further investigation, grid module is fragile by nature, so tests should be thorough.

./lib/python/pygrass/modules/testsuite/test_doctests.py

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-10-03-07-00/report_for_nc_basic_spm_grass7_nc/lib/python/pygrass/modules/test_doctests/index.html

pygrass: ‘VectorTopo’ object has no attribute ‘table’

My guess here is just a problem in a test itself.

./lib/python/pygrass/testsuite/test_doctests.py

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-10-03-07-00/report_for_nc_basic_spm_grass7_nc/lib/python/pygrass/test_doctests/index.html

gunitest: Only numeric column type is supported

Seems like the NC basic dataset has different column types then full NC. Needs further investigation. BTW, what about Datasets category for Trac?

./lib/python/gunittest/testsuite/test_assertions_vect.py

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-10-03-07-00/report_for_nc_basic_spm_grass7_nc/lib/python/gunittest/test_assertions_vect/index.html

Overall report:
Executed 128 test files in 0:27:25.642203. >From them 120 files (94%) were successful and 8 files (6%) failed.
Individual tests: Status: 94%, Tests: 1368, Successful: 1358, Failed: 9, Percent successful: 99%

neteler · October 4, 2015, 6:46am

On Sat, Oct 3, 2015 at 6:40 PM, Vaclav Petras <wenzeslaus@gmail.com> wrote:
...

Seems like the NC basic dataset has different column types then full NC.
Needs further investigation. BTW, what about Datasets category for Trac?

Done.

Markus

neteler · November 3, 2015, 11:31am

Hi Vaclav,

would it be possible to collect all "Failed" test in a single page?

Maybe as an extra entry here:
http://fatra.cnr.ncsu.edu/grassgistests/

thanks,
Markus

neteler · November 3, 2015, 11:50am

Hi again,

in addition, I feel that the Piemonte dataset on the server needs to be updated.

e.g.
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-02-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/vector/v.net/test_v_net/index.html
--> ERROR: Vector map <streets> not found
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-02-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.viewshed/test_r_viewshed/index.html
--> ERROR: Raster map <elevation> not found
(many more like that)

For our recent IRSAE course, we have prepared an updated version which
complies to the agreed upon map names:
https://trac.osgeo.org/grass/wiki/SampleDataset/Italy/Piemont
--> Download of draft location package
--> 7z compressed file

This dataset is to be further updated but already way closer to the
final version and it used the standard names.

Vaclav, could you please replace the Piemonte dataset on your test server?

thanks,
Markus

wenzeslaus · November 3, 2015, 5:52pm

On Tue, Nov 3, 2015 at 6:50 AM, Markus Neteler <neteler@osgeo.org> wrote:

Hi again,

in addition, I feel that the Piemonte dataset on the server needs to be
updated.

e.g.

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-02-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/vector/v.net/test_v_net/index.html
--> ERROR: Vector map <streets> not found

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-02-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.viewshed/test_r_viewshed/index.html
--> ERROR: Raster map <elevation> not found
(many more like that)

For our recent IRSAE course, we have prepared an updated version which
complies to the agreed upon map names:
https://trac.osgeo.org/grass/wiki/SampleDataset/Italy/Piemont
--> Download of draft location package
--> 7z compressed file

This dataset is to be further updated but already way closer to the
final version and it used the standard names.

Vaclav, could you please replace the Piemonte dataset on your test server?

Where to get the updated version? I already downloaded the one linked from
the page linked from Trac but now I'm looking at it and it is still the one
with old names.

thanks,
Markus

neteler · November 3, 2015, 6:09pm

On Nov 3, 2015 6:52 PM, “Vaclav Petras” <wenzeslaus@gmail.com> wrote:

On Tue, Nov 3, 2015 at 6:50 AM, Markus Neteler <neteler@osgeo.org> wrote:
.

Vaclav, could you please replace the Piemonte dataset on your test server?

Where to get the updated version? I already downloaded the one linked from the page linked from Trac but now I’m looking at it and it is still the one with old names.

Maybe F5? There is a new file from 2015.
Will check later if I put the wrong link.

Markus

thanks,
Markus

neteler · November 4, 2015, 11:30am

On Tue, Nov 3, 2015 at 6:52 PM, Vaclav Petras <wenzeslaus@gmail.com> wrote:

On Tue, Nov 3, 2015 at 6:50 AM, Markus Neteler <neteler@osgeo.org> wrote:

...

For our recent IRSAE course, we have prepared an updated version which
complies to the agreed upon map names:
https://trac.osgeo.org/grass/wiki/SampleDataset/Italy/Piemont
--> Download of draft location package
--> 7z compressed file

This dataset is to be further updated but already way closer to the
final version and it used the standard names.

Vaclav, could you please replace the Piemonte dataset on your test server?

Where to get the updated version?

My bad. Apparently I was in preview only, then trashed the browser
tab. Now there is really a new entry

https://trac.osgeo.org/grass/wiki/SampleDataset/Italy/Piemont
--> Download of draft location package
--> 2015: Piemonte (Italy) location with updated map names

sorry for the confusion,
Markus

wenzeslaus · November 5, 2015, 2:56pm

On Wed, Nov 4, 2015 at 6:30 AM, Markus Neteler <neteler@osgeo.org> wrote:

On Tue, Nov 3, 2015 at 6:52 PM, Vaclav Petras <wenzeslaus@gmail.com>
wrote:
> On Tue, Nov 3, 2015 at 6:50 AM, Markus Neteler <neteler@osgeo.org>
wrote:
...
>> For our recent IRSAE course, we have prepared an updated version which
>> complies to the agreed upon map names:
>> https://trac.osgeo.org/grass/wiki/SampleDataset/Italy/Piemont
>> --> Download of draft location package
>> --> 7z compressed file
>>
>> This dataset is to be further updated but already way closer to the
>> final version and it used the standard names.
>>
>> Vaclav, could you please replace the Piemonte dataset on your test
server?
>
> Where to get the updated version?

My bad. Apparently I was in preview only, then trashed the browser
tab. Now there is really a new entry

New version in palace and the maps are there

Old:
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-02-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/vector/v.net/test_v_net/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-02-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.viewshed/test_r_viewshed/index.html
--> ERROR: Raster map <elevation> not found

Current:
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-05-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/vector/v.net/test_v_net/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-05-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.viewshed/test_r_viewshed/index.html
AssertionError: v.info map=test_vnet layer=1 -t difference:
mismatch values (key, reference, actual): [('lines', 50069, 19568),
('nodes', 42136, 14466), ('points', 167, 282)]
...
WARNING: Region extent: north=1, south=1, east=2, west=698222432
WARNING: Region extent: rows=10660, cols=8160
WARNING: Viewpoint: row=584, col=12868
ERROR: Viewpoint outside of computational region

Finally we can focus on the real challenge!

https://trac.osgeo.org/grass/wiki/SampleDataset/Italy/Piemont
--> Download of draft location package
--> 2015: Piemonte (Italy) location with updated map names

sorry for the confusion,
Markus

neteler · November 15, 2015, 1:10pm

Hi,

I checked to Piemonte tests and saw a few issues while overall the
success rate is much better now:

#######
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.in.gdal/test_r_in_gdal/index.html

ERROR: Projection of dataset does not appear to match current location.
Location PROJ_INFO is:
name: Universal Transverse Mercator
proj: utm
...
Dataset PROJ_INFO is:
name: Lambert Conformal Conic
proj: lcc

... how should this be handled? The Piemont data are in UTM, so the
data in raster/r.in.gdal/testsuite/data/ do not fit since written for
NC.

#######
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.profile/test_profile_ncspm/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.viewshed/test_r_viewshed/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/scripts/r.import/test_r_import/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.watershed/r_watershed_test/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.slope.aspect/test_r_slope_aspect/index.html

--> also here the tests are NC specific.

Maybe they should be skipped in the Piemont test case?

#######
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/vector/v.what/test_vwhat_ncspm/index.html
ERROR: Vector map <elev_points> not found
--> this map only exists in location "nc_basic_spm_grass7", hence not
in Piemont (which is a clone of the full NC location).
I would suggest to replace that map with some other point map being
present in the full datasets.

#######
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/lib/python/pygrass/modules/test_doctests/index.html

File "etc/python/grass/pygrass/modules/grid/grid.py", line 85, in
grass.pygrass.modules.grid.grid.copy_mapset
Failed example:
    sorted(os.listdir(os.path.join(path, 'PERMANENT')))
Expected:
    [u'DEFAULT_WIND', u'PROJ_INFO', u'PROJ_UNITS', u'VAR', u'WIND']
Got:
    [u'CURGROUP', u'DEFAULT_WIND', u'MYNAME', u'PROJ_INFO',
u'PROJ_UNITS', u'VAR', u'WIND']

--> API change?

#######
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/vector/v.net/test_v_net/index.html
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/lib/python/pygrass/modules/grid/test_doctests/index.html

--> API change?

#######
http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/scripts/r.import/test_r_import/index.html

File "etc/python/grass/gunittest/checkers.py", line 264, in values_equal
precision = float(precision)
TypeError: float() argument must be a string or a number

--> not sure what's wrong here.

Markus

wenzeslaus · November 15, 2015, 10:22pm

On Sun, Nov 15, 2015 at 8:10 AM, Markus Neteler <neteler@osgeo.org> wrote:

#######

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.in.gdal/test_r_in_gdal/index.html

ERROR: Projection of dataset does not appear to match current location.
Location PROJ_INFO is:
name: Universal Transverse Mercator
proj: utm
...
Dataset PROJ_INFO is:
name: Lambert Conformal Conic
proj: lcc

... how should this be handled? The Piemont data are in UTM, so the
data in raster/r.in.gdal/testsuite/data/ do not fit since written for
NC.

#######

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.profile/test_profile_ncspm/index.html

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.viewshed/test_r_viewshed/index.html

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/scripts/r.import/test_r_import/index.html

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.watershed/r_watershed_test/index.html

http://fatra.cnr.ncsu.edu/grassgistests/reports_for_date-2015-11-15-08-00/report_for_piemonte_utm32_wgs84_grass7_stdmaps/raster/r.slope.aspect/test_r_slope_aspect/index.html

--> also here the tests are NC specific.

Maybe they should be skipped in the Piemont test case?

These will probably need to be disabled for NC. I had the support there but
then I drop it at one point to implement something more useful at that time
(I would have to look to commits what was the reason). The basic idea is
that the tests will state (probably in variable or in a comment) which
Location or kind of Location they want and then they won't be executed (the
command line API say which tests you want to run is there: --location-type
nc). What is Location and kind of Location is not completely clear but we
need to figure that out on the way. The biggest work is probably in the
aggregation scripts. I plan to look at it but so far I was just postponing
it, now it seems there is not much reason to postpone it more.

Doc with some ideas about how the problem should be approached:
https://grass.osgeo.org/grass71/manuals/libpython/gunittest_testing.html#data

Vaclav