[fitsbits] Abuse of EXTEND keyword

[Preben Grosbol]

On Thursday 23 August 2007 03:25, Tim Pearson wrote:
> My guess (it's a long time ago) is that there were FITS files out  
> there that had special records that did not conform to the rules for  
> standard extensions, so the EXTEND keyword was needed. Or perhaps  
> Grosbøl et al wanted to leave open the possibility of other uses for  
> the special records, which would be indicated by different header  
> keywords.

At the time, special records were used (most noticeable by random groups).
Thus, it was important to have a flag  (EXTEND=T) to indicate that the file
conformed to the new rules for FITS extensions.  Due to the wide usage
of random groups, special rules were introduced to allow coexistence but 
other less used types of special records existed.

This flagging function (as used with GROUPS=T in random groups) indicates
that all special records conform, as stated in the last part of the sentence.
Thus, EXTEND=F could mean one of two things: 1) the writer knew about
the new rules for extensions but did not care/conform, or 2) there are no
extensions (which however is a subcase of EXTEND=T).  As I said previously,
it also visually showed the user, by listing the primary header, that
information could be present in extensions.

During the introduction of the extension rules, the lack of EXTEND=T 
showed clearly that the writer did not know about the new rules. Now
almost 20 years later, this is not an issues.  We may ask if anybody
know of FITS files (possibly archival) which use their own 'special' 
special records.  Even so I see no major problem in making the keyword
optional butit's not a significant issue.

I prefer such flags (GROUPS, BLOCKED, EXTEND) as they, by nature,
just indicated an additional convention.  If one introduces an explicit
version, it may be too easy to make (too) strong changes.

Preben

PS: I do notice and appreciate the 'ø'.