[fitsbits] CDELTn

[Thierry Forveille]

> The only other point is that CDELTn were around before WCS,
Around before the WCS paper yes, but not around before WCS: a
minor source of confusion in this discussion has been that WCS
existed from the very begining of FITS (the WCS phrase may have
appeared later, not really sure, but the concept was there from
the begining), in their basic linear and orthogonal form. What
the WCS paper did was add support for distortion, projections,
but basic WCS is much older.

>> 3) If CDELTia isn't meaningful, as for a degenerate axis, then simply
>>    omit it.  It defaults to 1.0.
>
> Looks a sensible suggestion. Not using degenerate axes is perhaps even
> more sensibile :-)
>
Not at all :-) Degenerate axes are actually a most natural (if perhaps
initially slightly counterintuitive, for some) way of conveying position
information along the missing axes. One classical example is using
two degenerate axes to convey RA & Dec (or lII & bII, or whatever)
information for a set of individual spectra obtained on a grid,
where you can set CDELTn to the grid spacing, CRVALn to RA at grid
center, and CRREFn to the pixel position of the grid center relative
to current datum (i.e. minus the pixel position of current datum
in the grid). Very convenient, and commonly used for single-dish
radioastronomical spectra.