MATRIXNUMERIC with nxcals

[ydutheil]

Hi,

The nxcals part does not seem to handle the MATRIXNUMERIC datatype conversion. I saw this with the variable TD68.BTVDD.689339.B1/Image#imageSet.

Still, I can use the methods getMatrixLongValues and getMatrixDoubleValues on the Java object returned but I only see empty arrays.

What do you think ?

[rdemaria]

can you write a full example?

[ydutheil]

import pytimber
import dateutil

db = pytimber.LoggingDB(source="nxcals")

t1 = dateutil.parser.parse('2017-10-16 00:00:00.000')
t2 = dateutil.parser.parse('2017-10-19 00:00:00.000')

var = 'TD68.BTVDD.689339.B1/Image#imageSet'
d = db.get(var, t1, t2, unixtime=False)

timestamp, ds = d[var]

[test.getMatrixDoubleValues() for test in ds]

[test.getMatrixLongValues() for test in ds]

returns an array of <java array class [[D> and for the Long method an array of <java array class [[J>

[rdemaria]

You can use:

np.array([np.array(dd.getMatrixLongValues()) for dd in ds])

for the time being. I try to see what is going wrong in the code

[ydutheil]

I tried but that does not give the correct results, only the first value of each array is nonzero. However, that first value is correct.

import pytimber
import numpy as np

db_nxcals = pytimber.LoggingDB(source="nxcals")
db_cals = pytimber.LoggingDB(source="ldb")

t1 = '2017-10-16 00:00:00.000'
t2 = '2017-10-19 00:00:00.000'

var = 'TD68.BTVDD.689339.B1/Image#imageSet'

timestamp_cals, ds_cals = db_cals.get(var, t1, t2, unixtime=False)[var]

timestamp_nxcals, ds_nxcals = db_nxcals.get(var, t1, t2, unixtime=False)[var]

for d in ds_cals:
    print('CALS', d.sum())
for d in [np.array(dd.getMatrixLongValues()) for dd in ds_nxcals]:
    print('NXCALS', d.sum())

[rdemaria]

I enable matrixnumeric support in last commit 06aea6a, but then the problem seems more likely in the backport api. I contact nxcals team...

[rdemaria]

NXCALS team found an issue in the backport api

[ydutheil]

Hi,
now when I try to get the MATRIXNUMERIC data I get an empty array. Following with the debugger I find that dataset.isEmpty() is True on the function processDataset which returns an empty array.

Let me know what you think.

PS : I tried to run your example nxcals_backportapi.py but nxcals returns an array of java objects.

[rdemaria]

I confirm I receive the empty dataset as well...