While I'm in favor of both of these proposals, it seems to me
that they need a formal document giving the precise words
which are to be added to the FITS standard. Indicating that
these are currently appendices B.1 and B.2 is not sufficient.
These have phrases like: "This facility is still undergoing
trials ..." and similar caveats that need to be deleted. The
first few words of B.2 are fine for a proposal for some
new feature, but seem quaint in light of the fact that FITS files
have been using this feature for years (decades?).

B.2 has the phrase "This convention is optional." which is very
unclear to me. If this phrase were to be retained would it mean
that someone could use TDIMnn = '(2,3,4)' to mean something other
than what B.2 currently specifies.

Ideally this document would indicate precisely the new words
in the standard and where these
new conventions are to be inserted into the existing standard.

Finally and more substantively, Bill and I have discussed the
limitation of the variable length records convention to 4-byte integer
sizes and offsets. This may begin to chafe fairly soon. While 2 GB will
usually be OK for the sizes for individual cells of a variable length
column, it is already very easy to build tables where the offsets
need to be more than 2 GB. While we may not want to include
8 byte pointers in the current proposal it might be useful
to have some discussion of this now.

Regards,
Tom McGlynn

William Pence wrote:

ANNOUNCEMENT: START OF FORMAL PUBLIC COMMENT PERIOD