[fitswcs] inherit vs. concatenate

[Malcolm J. Currie]

> On Tue, 15 Jan 2002, Eric Greisen wrote:
> >  > I guess the time for such comments passed several years ago though!
> > 
> > A very long time ago.  This paper is 10 years old and the deadline for
> > its submission to the journals is approaching.  The purpose of the
> > paper is to achieve a data interchange format, not to write software
> > systems.  
> 
> Firsly, I agree that we've got to get the current proposals published soon
> now.

Yessir.  I'm trying to define WCS headers for our most recent
instrument (Michelle mid-IR) and the possibly imminent UIST
instrument.  We need the set of headers defined before we can
archive the data.

Both instruments are multi-mode (imaging and spectroscopy), and UIST
has an image slicing IFU for spectroscopy of a rotatable area, and a
cross-dispersed mode too.  I'm further constrained by a single ASCII
header-definition table per instrument.  Oh lordy.

I just about got up to speed on the WCS and issues by ADASS, but
following a torrid two months of other software work I found that I
had to start over.  Like Don, I was taken aback by the reintroduction
of CDELTn, though personally I prefer the modularity of the PC matrix
especially for the last two reasons listed in Paper I 2.1.1.  We
shouldn't have been surprised, as Eric did intimate in conversation
with Don and myself that the PC matrix would be the logical choice
and far superior.  I'll try to post some comments, suggestions, and
questions on the revised papers this week.  The main question for
the Michelle headers is whether or not the PC matrix is here to stay.

The splitting of the distortions from Paper II and made more general,
does at least in part address David's wishes.  It also mimics the way
Starlink tackled transformations even before the AST design David
mentioned.  While David's object-oriented goals are laudable, we
simply don't have the time or credibility to start the FITS WCS from
scratch.  If in time the WCS creaks under the strain, we could
consider a WCS-II rather than adding more and more complexity to the
existing version.

Malcolm