Rasql_Project() does not work properly with multiband coverages

[Bang Pham Huu]

try this Rasql query on test_irr_cube_2 (2 bands) and check the output multiple times, they will have different values on second band (sometime is the same, sometime is different).

rasql -q 'SELECT encode(project( c[1:2,19:23,0],
 "90369.95,5205426.80,102529.014,5265958.546", 
"EPSG:32633", "EPSG:4326" ), "csv") 
FROM test_irr_cube_2 AS c' --out file

see different values here ​https://www.diffchecker.com/FHlioQdK

If manage to fix this ticket then need to uncheck from known fails in test_wcs as these query returns non-consistent value.

77-subsettingCrs_4326_to_32633.kvp
78-subsettingCrs_32633_to_4326.kvp

[Dimitar Misev]

Submitted a patch that fixes some issues with project on multiband arrays.
Not sure yet if it's completely correct as a test on test_rgb2 gives a somewhat strange result.

@Bang once applied, can you test it on proper referenced rgb and compare to gdalwarp?

[Dimitar Misev]

@Bang once applied, can you test it on proper referenced rgb data and compare to gdalwarp?

[Bang Pham Huu]

It is not fixed yet, import test_irr_cube_2 from test WCST Import and run this WCS request

http://localhost:8080/rasdaman/ows?service=WCS&version=2.0.1&request=GetCoverage&coverageId=test_irr_cube_2&subset=E(90369.9573136,102529.014652)&subset=N(5205426.8043,5265958.54617)&subset=ansi("2008-01-01T02:01:20.000Z")&subsettingCrs=http://www.opengis.net/def/crs/EPSG/0/32633&format=text/csv

which is translated to Rasql query

SELECT encode(project( c[1:2,18:24,0], "85042.7273594,5204865.55794,105042.7273594,5274865.55794", "EPSG:32633", "EPSG:32633" ), "text/csv" ) FROM test_irr_cube_2 AS c

and the output with some strange values (2.406136e-38) is different from oracle when running this request multiple times
​https://www.diffchecker.com/gHDwKLjG

[Dimitar Misev]

Please make a comparison with gdalwarp for RGB georeferenced data.

[Bang Pham Huu]

Replying to dmisev:

Please make a comparison with gdalwarp for RGB georeferenced data.

Here ​https://www.diffchecker.com/OVWDl3Av is the comparison of a result from

http://localhost:8080/rasdaman/ows?service=WCS&version=2.0.1&request=GetCoverage&coverageId=test_world_temp&outputCrs=http://opengis.net/def/crs/EPSG/0/3857&format=image/tiff

and gdalwarp on a 2D GTiff file with 3 bands (source CRS: EPSG:4326, target CRS: EPSG:3857).

[Dimitar Misev]

This has been fixed in #1913

And in the systemtest the test cases is indeed marked as FIX (= should be removed from known_fails):

test.sh: 128/210   FIX   .27s   078-subsettingCrs_32633_to_4326.kvp

The other one, 77-subsettingCrs_4326_to_32633.kvp doesn't seem to exist anymore.