[wfc] CHECKSUM Proposal

[Eric Greisen]

William Pence writes:
 > Eric Greisen wrote:
 > > 
 > > The FITS community also needs to look at another issue - whether all
 > > of the "conventions", even those that are widespread, ought to be
 > > "standards".  If something is a standard, then most widespread
 > > software systems ought to support it.  If it is a convention, we have
 > > a greater leeway.
 > 
 > Conventions that are widely used and that provide a significant capability
 > to FITS need to be added to the standard to preserve and document them for
 > other users.  If this isn't done then there is a danger that the precise
 > details of the convention will be forgotten or misinterpreted by others,  or
 > that other duplicate conventions will be invented and used by people who are
 > unaware of the other existing convention.  

     My convention questions was intended to be more philosophical.
There would have to be a place in which conventions get registered and
documentations maintained to avoid the ignorance/duplication to which
you refer.

 > 
 > Don Wells wrote:
 > >
 > > However the current form of the proposal contains only the
 > > technical details, and so it won't help me in the process of
 > > persuading the archive people to implement CHECKSUM.  In this respect
 > > the 1995 Seaman and Pence paper was better. Does the proposal
 > > presented to the WFC supercede the 1995 paper?  If 'yes', then I
 > > recommend that the 1995 paper be upgraded so that I can use it.
 > 
 > The current proposal is a more concise and reorganized version of the 1995
 > paper. (It is 7 pages long compared to 21 pages in the 1995 paper).

     We can vote on only one paper.  I am voting NO on the 7-page
paper on which we were asked to vote because it is not in a shape to
stand on its own.  I suspect that around 1-2 pages of introduction
and practical usage hints would be enough to fill it out.  I went back
to the NOST FITS Glossary and was still confused about the meaning of
HDU - one sentence to clarify the question I asked would avoid the
whole issue.

     I am not opposed to the proposal to have checksums included.  It
seems like a good idea, although we get very few if any reports of
failure these days on the 40+ Megabyte AIPS tar ball.  Don's question
about how often this is checked and found to prevent error is relevant
to those of us who would have to go to a lot of work (and cost to the
user) to implement the concept.

      What I am opposed to is the current presentation of the
proposal.  I have spent more than a year FTE on the WCS proposals to
make them not only correct but also fully readable and professional in
appearance as well as content.  And the committees never take them up
despite statements that they will.  Given that, why should I now vote
for some other proposal that does not meet the professional standards
for a standards proposal?

 > Everything in the 1995 paper is still valid.  The only technical difference
 > is that the current proposal does not explicitly reserve the CHECKVER
 > keyword for possible future use;  we decided that the need for this keyword
 > was so uncertain that there was no point in reserving it now.   Essentially
 > everything in sections 1, 2, 3, and 5, of the 1995 paper has been retained
 > in the current proposal.  Only section 4 of the 1995 paper has been omitted,
 > which briefly discusses a variety of other checksum schemes such as CRCs,
 > message digests, digital signatures, and error correction algorithms.   This
 > section was removed from the current proposal mainly because the discussion
 > of alternate schemes is somewhat of a digression and is not strictly
 > relevant to the CHECKSUM proposal itself, and it seemed inappropriate to ask
 > the FITS committees to vote on this 'off-topic' section.

Eric Greisen