Unhandled exception thrown in AveragedCrossspectrum constructor

If we create two lightcurves and then create an AveragedCrossspectrum without specifying the segment size. The following exception gets thrown: TypeError: unsupported operand type(s) for /: ‘NoneType’ and ‘float’.

The full traceback:

  File "\<stdin\>", line 1, in <module>
  File "\~/stingray/crossspectrum.py", line 541, in __init__
    Crossspectrum.__init__(self, lc1, lc2, norm, gti=gti)
  File "\~/stingray/crossspectrum.py", line 137, in __init__
    self._make_crossspectrum(lc1, lc2)
  File "\~/stingray/crossspectrum.py", line 606, in _make_crossspectrum
    self._make_segment_spectrum(lc1, lc2, self.segment_size)
  File "\~/stingray/crossspectrum.py", line 573, in _make_segment_spectrum
    dt=lc1.dt)
  File "\~/stingray/gti.py", line 597, in bin_intervals_from_gtis
    nbin = np.long(chunk_length / dt)
TypeError: unsupported operand type(s) for /: 'NoneType' and 'float' 

[haroonrashid235]

@matteobachetti So for now, best solution is to make segment_size a non-optional parameter. That way user will at least have an interpretable error message.

[matteobachetti]

@haroonrashid235 @omargamal8 I would rather force the user to specify it. I think that this is a parameter that users should choose explicitly and carefully. Also, it should not be separated from the bin time! We might easily create memory overflows or useless results. We might set a default value based on a certain number of bins, and check that we don’t go further than the GTIs in the observation… In summary, it’s more complicated than it seems, and for now I would just tell the user to act responsibly 😄

(@dhuppenkothen: what do you think about this?)