🚀 go-pugleaf

In article <1738@opus.NMSU.Edu> sitze@nmsu.edu (Richard Sitze) writes:

   ".... Someone else pointed out possible prior art in the form of a
   standard data format used by observatories, and I'm sure that quite
   often they run into the situation where Scheme X is translated into
   Scheme Y, and then used at the 'destination' by translating into
   Scheme Z..."

In January 1990 the FITS community activated an agreement to permit SP
and DP floating point in interchange. We adopted IEEE, so that means
that now a VAX writing a FP FITS file does VAX-->IEEE (and it does the
byte swap, of course!) and then when a 370 reads it it does IEEE-->IBM
(but doesn't need to do a swap, of course). FITS FP interchange
between diverse architectures works just fine.

   "...My only complaint is against those who feel that the idea is to
   OBVIOUS.  Yes, it is simple and a beautiful solution to the problem.
   But how many figured it out for themselves??..."

Those of us who negotiated the FITS Agreement in 1979 did indeed
figure it out for ourselves, as "lessons learned" from trial and
error. You see, there is another "obvious" alternative convention
which you did not mention: data type tagging.  We built two different
operational prototype formats during 1977-78 to test our ideas. Both
formats demonstrated interchange between independent implementations
on different architectures, a non-trivial achievement for 1978. One
format had the canonical data type and data order convention that we
ultimately used for FITS, while the other format had codes embedded in
the byte stream to indicate the coding convention, 1s-complement,
2s-complement, sign-plus-magnitude, IBM, CDC, etc. The idea of this
second design was "something for everybody", which would have the nice
political advantage that nobody (e.g. a VAX owner) would feel like a
second-class citizen, but the technical evaluation of the experiment
was "YUUCCH!".  So in the final FITS negotiation we decided to adopt
canonical types with a canonical byte order.

--

Donald C. Wells             Associate Scientist        dwells@nrao.edu
National Radio Astronomy Observatory                   +1-804-296-0277
Edgemont Road                                     Fax= +1-804-296-0278
Charlottesville, Virginia 22903-2475 USA            78:31.1W, 38:02.2N