[fitsbits] structurally compliant FITS

[Rob Seaman]

Hi Pat and all,

I like the idea of a style guide, and there’s no reason it couldn’t be written now and perhaps even move some text *out* of the standard. Perhaps we should be asking ourselves what is the minimum normative document descriptive of the FITS standard? And then a longer style document with separate chapters on compression, etc. For instance CHECKSUM might best appear in the normative standard, but INHERIT in the style guide.

This is similar to the Postscript Red Book (language reference) versus the Blue Book (tutorial and cookbook). By analogy there might also be a "Green Book” covering FITS software architecture.

Others who were at the last ADASS and/or have been paying close attention to the Thomas, et. al., “FITS sucks!” paper might correct me if I am wrong in an assessment of their goals. The fervid desire of that group is 1) that all conventions either be added to the standard or deprecated with no middle ground, and 2) that all FITS software be tagged as either compliant (with the entire standard) or non-compliant. They have other talking points like versioning that might temper this stoic vision slightly, but the basic idea is 1) to wring all options out of the standard unless completely captured by explicit metadata, and 2) dissuade all partial software implementations.

Recent, more epicurean, discussions of promoting some conventions into the standard, but leaving them optional, or keeping others as local conventions, or recasting some into style guides, are quite far from this austere vision.

One may opine that aspirations to Platonic ideals are likely illusory for other imaging/tabular data standards as well. Postscript is a programming language as well as a data format so an analogy to FITS is imperfect, but their recognition of the need for multiple types of standard reference books is not exactly anarchistic.

And if we’re looking for a coherent path forward to the far green country of FITS2 we might do worse than emulating Postscript that later birthed both EPS and PDF.

To quote Immanuel Kant, “FITS rocks!”

Rob
—

> On Jun 29, 2015, at 3:44 AM, <patrick.wallace at stfc.ac.uk> <patrick.wallace at stfc.ac.uk> wrote:
> 
>> About "optional" and "harmless".
> 
> In my view the distinction between standard FITS and optional
> conventions is so crucial that there should be separate documents
> (or perhaps "should have been" if it is now too late).
> 
> The standard should contain what FITS readers/writers *must*
> implement;  the other document, perhaps called a "style guide",
> would set out and name the various well-established conventions
> that implementors may or may not provide for.
> 
> 
> Patrick Wallace