Logic error in refl gui

[Owen Arnold]

The following is a bug introduced during last iteration's changes:

line 634 in refl_gui.py:

if isinstance(theta, numpy.float64):
        th = theta.size # This is wrong!
    elif not isinstance(theta, float):
        th = theta[len(theta) - 1]
    else:
        th = theta

The logic should be cleaned-up anyway.

[Keith Brown]

Refs #9037 Logic Error fixed

This was purely down to my own bad understanding of how numpy worked. It should now work as expected.

I also removed a debugging print statement as that was missed in another ticket.

Changeset: f13e8d15ff15fe5338fa639fcff053da0cb1df9d

[Keith Brown]

Refs #9037 Streamlined check for an array

found out that groupget was already doing a check (it seems i did it and forgot about it) so the check in calcres wasn't required.

The array accessing in groupget has also been streamlined

Changeset: 718afe437517bd6c1ec77c021ad08bdf11c2ca13

[Keith Brown]

To tester:

You won't need to start the gui for this test, as the functions i edited in the python script are public so these can be tested in the python window.

This tests calcRes() and groupGet() with numpy.ndarray returning properties. groupGet() handles all the log fetching and if a log is a numpy.ndarray it'll return the last element of it.

You will need INTER00013460.nxs which is found in the ISIS sample data: ​http://download.mantidproject.org/download.psp?f=/SampleData/SampleData-ISIS.zip

Run this script:

from ui.reflectometer import refl_gui
from time import gmtime, strftime,mktime
from numpy import *
#Load a reflectometry workspace
w=Load("INTER00013460.nxs")
#change the theta log inot a time series property so it's actually a numpy array
tlog=kernel.FloatTimeSeriesProperty("theta")
for i in arange(25):
   tlog.addValue(strftime("%Y-%m-%d %H:%M:%S", gmtime(mktime(gmtime())+i)),1.0*i*i)
#add the fudged log
w.mutableRun()['theta']=tlog
#check the refl_gui script
logVal = w.mutableRun()['theta'].value
gg_return = refl_gui.groupGet(wksp=w, whattoget='samp', field='theta')
print "LOG VALUE" 
print logVal
print "LOG VALUE TYPE", type(logVal)
print "REFL GROUPGET RETURN" , gg_return
print "GROUPGET RETURN TYPE", type(gg_return)
print "VALUE COMPARISON", logVal[-1] == gg_return
refl_gui.calcRes(w)

the test should output the following:

• 'Log value type' should be a numpy.ndarray
• 'groupget return type' should be numpy.float64
• 'Log value's last element and 'refl groupget return' should be the same
• 'value comparison' should be true
• calcres will output 3 values, theta should match 'refl groupget return'

[Owen Arnold]

Merge remote-tracking branch 'origin/bugfix/9037_refl_numpy_fix'

Full changeset: ​65c38c2051b86767cbe48140364192550f04216b

[Owen Arnold]

Good use of python for a verification step by the way. More of this in future please.

[Nick Draper]

Moved to patch release 3.1.1

[Keith Brown]

Refs #9037 Logic Error fixed

This was purely down to my own bad understanding of how numpy worked. It should now work as expected.

I also removed a debugging print statement as that was missed in another ticket.

Changeset: 0294354b38f02de4cc939a709edc83977c7c97ed

[Keith Brown]

Refs #9037 Streamlined check for an array

found out that groupget was already doing a check (it seems i did it and forgot about it) so the check in calcres wasn't required.

The array accessing in groupget has also been streamlined

Changeset: 3a0ec2626bb507da9891d426689073d2e791e11c

[Russell Taylor]

The commits above come from putting the work in this ticket into the patch release branch.

[Stuart Campbell]

This ticket has been transferred to github issue ​9880