[fitsbits] TDIMn

[Clive Page]

I would like to support the message from Jonathan McD: X-ray astronomy
tables do indeed make extensive use of the TDIMn and variable length array
facilities.

On Mon, 30 Aug 2004, William Pence wrote:

> B.3  The "Substing Array" convention

> I'm not aware of any software packages or any FITS data files that support
> or use the full form of this proposed convention, as in these examples:

> TFORMn = '30A:SSTR10'      / fixed length 10-char substrings

I have used this convention myself for FITS files I have generated and
used internally, but I am not sure that it has been used much in files
distributed externally.  If one wants a vector column of a numerical data
type one simply uses e.g. 5E rather than 1E, which is simple to do and
understand.  The same facility for strings needs the fixed-length
substring notation, otherwise you just get an array of chars.

I suspect that vector columns are not widely used for two reasons: (1)
ignorance, (2) the false assumption that not many utilities can handle
them.  This leads to results which are not always very elegant, for
example the 1XMM catalogue of serendipitous sources from XMM-Newton has
been circulated as a table with 394 columns.  If vector columns had been
used for sets of related items (e.g. for the 3 telescopes on board), there
would have been many fewer columns, and the naming could have been simpler
and less obscure.

As it is, however, if one wishes to ingest a FITS table into a relational
DBMS, then the vector-free form is easier to use, since few DBMS support
vector columns.

It is hard to know the right balance here: on the whole I think that
vector columns in FITS files are useful and here to stay, so they should
be incorporated into the Standard.  It is something of an anomaly that
FITS has no notation of a character "string" only of an array of char
(perhaps following the primitive notation of the C language rather than
the more sensible ones of say Fortran or Python).  On grounds of
consistency with the other data types, therefore, it seems to me a good
idea also to incorporate the substring notation into the Standard.


-- 
Clive Page
Dept of Physics & Astronomy,
University of Leicester,
Leicester, LE1 7RH,  U.K.