[fitsbits] updates to the FITS standard document

[Lucio Chiappetti]

On Tue, 23 Jun 2015, Douglas Tody wrote:

> The INHERIT convention originally came from IRAF ...

First of all I clarify that the purpose of the Public Comment Period is 
just to assess the feeling of the community. It might well be that the 
outcome is NOT to move SOME of the conventions to the voting stage.

In this respect the convention technical team had just the task of seeing 
how easy it was to rewrite some of the existing conventions in 
prescriptive way fit for the standard document. The assignment of a 
particular convention to a particular editor did not necessarily imply a 
strong support.

In this respect by chance I "did" INHERIT because it was a simple one to 
do, and I had a neutral approach when this was discussed years ago. Let us 
say I considered it harmless, but I haven't made use of it myself in 
homegrown code.

Please redirect suggestions to NOT include a given convention (or not pass 
to vote) in the thread of the specific convention.


> There are additional issues with what to do if a single extension is 
> extracted from the MEF ...

> Hence one solution is merely to avoid relations and inheritance and 
> merely duplicate the information in every FITS entity.

All this is true. In general the arrangement of a MEF is something which 
would benefit "regulation". I've never been very keen in long complex MEF, 
unless they are just sort of tar files (or FITS Archive Files). We have 
been discussing in the other Technical Team (the one about new feature) 
ways of improving metadata, but did not come yet to a conclusion.


> On this general issue of incorporating FITS conventions into the FITS
> standard, I think the main issue is, are we merely introducing the major
> conventions in the standard document as optional extensions, or are we
> recommending these as integral components of the standard?  T

> Probably some belong in one category or the other.

The difference is not so strong, at least for the conventions considered 
so far. The difference between a plain convention, a "registered 
convention" and a "standard convention" is

- a plain convention SHOULD be legal FITS but there is no guarantee,
   documentation and definition relies with the original author, usage
   is fully optional and/or confined to specific projects

- a registered convention IS legal FITS, and has a COMPLETE documentation
   (that's what approval of the registration is for). It might have general
   nature, or be specific of a project or subcommunity (but can interest
   other communities with similar goals). Essentially a registered
   convention means

   'if you want to do this, here is how we do it. If you need to do the
   same, and like how we do it, you can do the same'

- a convention in the standard in general remains optional, but means

   'if you need and want to do this, you shall do it this way'

In this respect

  tiled compression and CHECKSUM are optional, but if one needs to
  compress, or needs to have a checksum, they provide a good way to
  do it. Personally I would never write homegrown s/w to do it, but
  I will be glad to use existing tools doing it

  header space and Green Bank are "mere" usage conventions (there are
  no signature in the file saying it adheres to them) and as such
  again optional

  INHERIT and min/max are also optional (they are "signed" by the
  presence of the relevant keyword/s). I regarded both of them as
  harmless (and surely this can be the case for min/max)

  CONTINUE is probably one of the most useful ones. It is optional in the
  sense that if one does not need long keyword values, one does not use it.
  It was designed to be harmless (dealt with as commentary) but the
  code changes to fully support it as part of the standard are limited.

-- 
------------------------------------------------------------------------
Lucio Chiappetti - INAF/IASF - via Bassini 15 - I-20133 Milano (Italy)
For more info : http://www.iasf-milano.inaf.it/~lucio/personal.html
------------------------------------------------------------------------
Do not like Firefox >=29 ?  Get Pale Moon !  http://www.palemoon.org