WCPS1.5_Difference matrix values of bands calculation from WCPS 1.0 (Need to update oracle test)

[Bang Pham Huu]

wcps test "04-vegetation_index.test.out" with WCPS query

for c in ( rgb ) return encode((unsigned char)
( ((c.0 / ((float)c.0 + c.1))-(c.1 / ((float)c.0 + c.1))) > 0.6 ) * 255, "png", "nodata=0" )

is translate in Rasql by WCPS 1.0

select encode((char)((((((((((c).0)/((float)((((c).0)+((c).1))))))-((((c).1)/((float)((((c).0)+((c).1))))))))>(0.6)))*(255))),
 "PNG", "nodata=0;xmin=0.0;xmax=399.0;ymin=0.0;ymax=343.0;crs=OGC:Index2D")
 from rgb AS c where oid(c)=45057

and Rasql by WCPS 2.0

SELECT encode((char)  
(  (  ( c.0 /  ( (float) c.0 + c.1 )  ) 
 -
  ( c.1 /  ( (float) c.0 + c.1 )  )  )  > 0.6 )  * 255,
 "png" , "nodata=0") FROM rgb AS c

the result is different, view here ​https://drive.google.com/a/fimo.edu.vn/folderview?id=0B_VL2-SVlKPHU1VMSi1iXzgzNTA&usp=drive_web (oracle is WCPS 1.0, output is WCPS 2.0)

due to the matrix calculation between coverage bands which is different between WCPS 1.0 and WCPS 2.0

[Dimitar Misev]

Did you make a mistake in copying the rasql queries? Because both queries seem to be exactly same, please check.

[Bang Pham Huu]

Thanks Dimitar, I've posted wrong WCPS 2.0 Rasql query.

[Dimitar Misev]

I'm still not sure what's the difference between the queries, can you point it out?

[Bang Pham Huu]

Dimitar, view images in ​https://drive.google.com/a/fimo.edu.vn/folderview?id=0B_VL2-SVlKPHU1VMSi1iXzgzNTA&usp=drive_web you will see that WCPS 2.0 has more pixel has values than WCPS 1.0

[Dimitar Misev]

No I mean the rasql queries, not the outputs. The WCPS 1 query doesn't seem like it came out of petascope?

[Bang Pham Huu]

It is different in syntax while WCPS 1.0 has a lot of (((())) but I don't know about what do you mean?

[Dimitar Misev]

Is this exactly the query that petascope sent to rasdaman:

and Rasql by WCPS 2.0

SELECT encode((char)  
(  (  ( c.0 /  ( (float) c.0 + c.1 )  ) 
 -
  ( c.1 /  ( (float) c.0 + c.1 )  )  )  > 0.6 )  * 255,
 "png" , "nodata=0") FROM rgb AS c

[Dimitar Misev]

Ok in this case WCPS 2 translates correctly, as WCPS 1 messes up something with the parentheses (there should be no parentheses in the bold part):

• select encode((char)(((((( c.0/((float)(c.0+c.1))))-((c.1/((float)((c.0+c.1)))))))>0.6))*255
• SELECT encode((char)( ( (c.0/((float)c.0+c.1) ) - (c.1/((float) c.0+c.1))) >0.6) *255

[Bang Pham Huu]

Supper, then someone need to fix in WCPS1.0 ? or let it there.

[Dimitar Misev]

No let's just update the correct oracle.

[Bang Pham Huu]

but have to wait until WCPS 2.0 is available or system test will fail with WCPS 1.0, then the ticket here is just update oracle for case "04-vegetation_index.test.out"

[Vlad Merticariu]

This is related to ​http://rasdaman.org/ticket/589

[Dimitar Misev]

Replying to bphamhuu:

but have to wait until WCPS 2.0 is available or system test will fail with WCPS 1.0, then the ticket here is just update oracle for case "04-vegetation_index.test.out"

I think it would be best if you make a single patch switching to WCPS 2 in the systemtest and fixing all these tickets.

[Dimitar Misev]

Replying to dmisev:

I think it would be best if you make a single patch switching to WCPS 2 in the systemtest and fixing all these tickets.

Actually I think this is not such a good idea

How about temporarily adding an option to the test.sh script, to enable switching to the new WCPS in the systemtest, e.g. ./test.sh --wcps=1.5? Then each ticket can be fixed one by one, and the oracle updates can be done last at the end.

[Vlad Merticariu]

Bang is working on a different branch anyway, so on that branch he could enable systemtests to use wcps 1.5 and then fix 1 ticket at a time (1 fix = 1 patch).

[Dimitar Misev]

Ah right perfect.

[Bang Pham Huu]

the patch was accepted, close ticket, thanks.