Re: [guppi-list] Changes to make egcs work

[Jon Trowbridge <trow emccta com>]

On Wed, Feb 10, 1999 at 04:40:50PM +0100, Asger K. Alstrup Nielsen wrote:
> Take an existing DataSet derivate that does not support
> missing data, and *extend* it into one that does,
> and that still support the extra functionality that the
> original had.  Pretty neat.

That is neat.  I haven't had a chance to look at the code yet, but one
question/observation.  I was thinking about how to add censored values
(i.e. missing data) to a RealSet a while ago, and the thing which
tripped me up were the semantics of dealing with the gaps.  For
example, say you sort a RealSet w/ holes.  Where do the holes go?
Or if you rank-transform the data?  

There might be a standard statistics convention that we should follow
with regards to these questions.  However, I don't know what they
are...