[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]Author Index][Subject Index]
Klaus Banse wrote:
> P.S. That leaves us with the question of what to do with such a huge SELIDX
> descriptor resulting automatically from the SELECT/TABLE command.
> Since the logical expression in that command can fill up all paramters
> P2, ..., P8 there is no "space" for a YES/NO flag for the descr. creation.
> Should we just put in a limit of, say, 100 000 elements or leave it to the
> discretion of the user to later on delete that descr. if it's not needed?
I would very strongly argue against any explicit or implicit limitation on
table or descriptor size. Somebody may want to work with SELIDX, whatever its
length (it has to have a purpose). As the problem occurs when the many
descriptor elements are converted to FITS header lines, OUTDISK/FITS would be
the place to check whether a descriptor is longer than a certain limit, issue
a warning that this will take lots of time and space to write out as .fits
file, and ask the user what to do. If she doesn't want all the descriptor
elements saved in the fits file, she can still delete by hand.
IMHO, it is best just to make the user aware of this circumstance and let
everyone decide what to do about it, rather than guessing what might be best
and leaving no choice.