odd dim_of behavior for time axis in BES module

[drsmith48]

@ktritz @hyyuh When you get a chance, checkout branch 'bes' and run the following lines:

nstx = Machine(shotlist=141000)
nstx.s141000.bes.ch_01.plot()

You'll see that the time axis in the plot window is array indices, and some diagnostic print()'s in Machine._get_mdsdata() yield:

start: dim of
(0, 3078015)
[0 1 2 3 4 5 6 7 8 9]
end: dim of

Any idea why Machine._get_mdsdata() and bes.xml fail to grab proper time data?

[ktritz]

Hey Dave, I can't find your time data on the MDS tree. How did you get to it from Matlab and/or IDL?

[drsmith48]

Matlab:

timetag = 'dim_of(\activesp_raw::be_dt32_01:input_01)';
time = mdsvalue(timetag);

IDL:

timetag =  'DIM_OF(\ACTIVESP_RAW::BE_DT32_01:INPUT_01)'
time = mdsvalue(timetag, status=stat1)

[ktritz]

Yeah, for some reason doing that through python doesn't produce the time array, it just gives the indices. Wonder if this is a problem with the MDS client. You're not dividing by your sampling frequency or something later on to get your time array, are you?

[ktritz]

Also, check out my bes.xml in master, you can see how I used the 'range' tag to generate your channels. I also stuck your plot method as the default 1D plot in container which is inherited by all signals because I wanted to use it for USXR too. We may stick default methods somewhere else in the future, but you can use the default for your BES if you want.

[drsmith48]

@ktritz nope, I'm not modifying the time array. thanks for the range edit. I saw that in usxr, but I didn't know how to start the range at 1.

[ktritz]

Ok, I'll double check to see if it's working for the usxr, which uses the same dim_of language I think. This might be a python client thing, but you would think that this would have been caught and fixed by now if that were the case.

[drsmith48]

I get the same index list for the time axis with usxr data. it seems dim_of='0' does not return the proper time axis. ip, vloop, and mpts all have separate signal nodes for the time axis and do not use dim_of='0'.

[ktritz]

Ok, sounds like we need to hit up the mdsplus mailing list to find the answer.

[ktritz]

weird, 'dim_of()' seems to work now, and I could have sworn it didn't before. I'll try modifying the code to see if it fixes it.

[ktritz]

yup, tested working for both bes and usxr on the PC and linux.