test_read_fits_spec.py, the first two test cases failing

[shaileshahuja]

https://github.com/astropy/specutils/blob/master/specutils/io/tests/test_read_fits_spec.py#L20
iraf wave array:
[4344.5500522758402, 4344.59028429296, 4344.6305137707004, 4344.6707407088797, 4344.7109651072797]
fits dispersion array:
[4344.5098177195632, 4344.5500522758412, 4344.5902842929554, 4344.6305137707031, 4344.6707407088788]

The (n+1)th element in fits dispersion array is the same as the nth element in the iraf wave array.
@wkerzendorf @nhmc

[shaileshahuja]

@wkerzendorf @nhmc this issue needs to be resolved before #69 can be merged. Is the problem with the IRAF data files, or the WCS functions?

[nhmc]

I believe the problem is with the WCS functions - they assign the first pixel a value of zero, (instead of one, which is what IRAF does). There's a keyword to the WCS method .wcs_pix2world() which accepts 0 or 1 to indicate the indexing convention. So I think that's all that needs to be changed.

[nhmc]

It looks like the arange at line 263 in https://github.com/astropy/specutils/blob/master/specutils/spectrum1d.py needs to be changed to accept a starting index of either 0 or 1. I would add a new keyword to Spectrum1D.__init__ to indicate either a 0 or 1 indexing convention. This would set an attribute to be checked in Spectrum1D.dispersion to see whether arange should start with 0 or 1.

@wkerzendorf, do you think this is the best approach?

[wkerzendorf]

What @nhmc says, is very likely the cause of the problem. The indexing in IRAF starts with 1 (maybe because FORTRAN uses this idea). I think, however, that we should stay with 0-indexing throughout specutils. That will avoid any pitfalls. As a step forward we need to make sure that when we read-in WCS from fits files to subtract 1.

[shaileshahuja]

This issue will be fixed with #72