[fitsbits] (no subject)

[LC's No-Spam Newsreading account]

I open a separate thread.

On Fri, 13 May 2005, Mark Calabretta wrote:

> While we're counting bits, I note that the precision of keyvalue data
> types is essentially undefined.  [...]
>
> PSRTIME =       88402115102000 / [us] Time since T0
> PULSENO =           9465023873 / Pulsar pulse number since T0  

It is true that header keyword are not strongly typed. That's somehow a 
flaw of FITS as working format (in fact I've used for some applications 
a format of my own in which header keywords were stored as binary data, 
or binary data arrays (arrays of keywords are another thing I miss), 
with :

 - a 8-character FITS-compliant name
 - a 1-byte type specification
 - a 1-byte length word (number of items in number or array or string
   length)
 - n binary values of appropriate type, where the types I supported were
   INTEGER*4, REAL*4, REAL*8 and CHARACTER*n. I supported arrays up to
   255 elements, and strings up to 69 characters (this latter limit to
   be FITS compliant !)

If a keyword is typed, the application program can query the type, 
and/or store the entire header in a memory buffer (the way I used in 
Fortran) a data structure (the way I decoded my files into IDL)

Since FITS keywords are not strongly typed, the burden of reading them 
into a variable of appropriate type is left to the application writer.

I do not see any easy way of changing this in the "once FITS forever 
FITS" rule.

Lucio Chiappetti

-- 
----------------------------------------------------------------------
nospam at mi.iasf.cnr.it is a newsreading account used by more persons to
avoid unwanted spam. Any mail returning to this address will be rejected.
Users can disclose their e-mail address in the article if they wish so.