Further questions about the standard

[Thomas A. McGlynn]

In response to Bill's question about lower case values from
the TDISPn keywords: I agree with hime that the standard
requires uppercase but I suppose readers can do whatever
they want with non-standard files.

However, I think there is a problem with another clearly
stated rule for FITS  that only the first 8 characters
of a character string value can be significant in decoding
the values in the FITS file.  I believe the following two
columns for a binary table are illegal.

TFORM1 = '16000000B'
TDIMS1 = '(4000,4000)'
TFORM2 = '1000I'
TDIMS2 = '(10,10,10)'

In the first case both the TFORM and TDIMS keyword
are invalid while in the second only the TDIMS is too long.

The first row might be a little unusual, but is certainly
feasible with current detectors.  The second is rather common.

I think that FITS binary tables need to support columns like these.
Am I right that they are illegal?  If so what should be about
this?

		Regards,

		Tom McGlynn
		tam at silk.gsfc.nasa.gov

P.S., On an unrelated topic, I've been revising Java FITS classes.
I'd be curious to know if there are any forseeable needs for FITS
to support long (i.e., 8 byte) integers now that this data type
is becoming more commonly available.