Opened 6 years ago
Closed 6 years ago
#1789 closed defect (fixed)
WCPS_Fit output geo bounds from WCPS trimming should regard to negative grid origin
Reported by: | Bang Pham Huu | Owned by: | Bang Pham Huu |
---|---|---|---|
Priority: | critical | Milestone: | 9.7 |
Component: | petascope | Version: | development |
Keywords: | Cc: | Dimitar Misev, Vlad Merticariu | |
Complexity: | Medium |
Description
From problem in Rasdaman user group: https://groups.google.com/forum/#!topic/rasdaman-users/X9WrCeZhnyM
The problem is a coverage was imported with negative grid domain (map mosaic), e.g:
rasql -q 'select sdom(c) from l2a_b02 as c' --out string rasql: rasdaman query tool v1.0, rasdaman 9.6.0. Opening database RASBASE at localhost:7001... ok. Executing retrieval query... ok. Query result collection has 1 element(s): Result element 1: [0:30977,-40002:10979,0:2]
with its geo domains:
<lowerCorner>300000 4790220 "2017-11-22T10:53:41.000Z"</lowerCorner> <upperCorner>609780 5300040 "2018-03-22T10:50:21.000Z"</upperCorner>
And user does a trimming subset in WCPS, e.g:
$b[E(365000:385000), N(4820000:4840000)
However, from the source code http://rasdaman.org/browser/applications/petascope/petascope_main/src/main/java/petascope/wcps/metadata/service/SubsetParsingService.java#L313, this always convert to grid bound with origin = 0
, hence the translated grid bounds for the N axis
is: 46004:48003
which is much greater than the imported grid domain -40002:10979
and it causes other trimming/slicing (e.g: clip()) on this axis will be wrong grid coordinates.