Re: Issue 3: (STANDARD_MEASURES)

[hardie@xxxxxxxxxxxxxxxxxxxxxx (Ted Hardie)]

I would like to hear a bit more about why we should default
to SI, rather than metric, in this context.   Given the ease
of conversion among metric units of the same type,
limiting the metric units to just the "base" units of meters,
kg, s, doesn't seem to be a big win.  There are already a number
of systems which would seem to prefer things like ppcm ,
and I don't quite understand yet why we should make it less
intuitive for those users.

I raise it in this specific context, because the text below
seems to go back to the idea of "pixels per" with a unit
assignment of meters/centimeters/inches at the time
of use.  I strongly prefer a single assignment at registration,
but I'm not sure if I follow the arguments below on how
that would affect when an error condition is produced.  In
particular, I don't understand when someone would
want to register something with indicating a default
unit; are there examples for us to chew on?
		regards,
			Ted Hardie
			NASA NIC






---End of forwarded mail from Chris Yarnell <cyarnell@coredump>





> Al,
>
> I basically agree with what you say.  Especially the final paragraph.
>
> I am copying your entire message to the list because I don't think it got
> there first time because of a typo in the address, tho it was clearly
> intended to.
>
> #g
>
> At 12:34 28/05/98 -0400, Al Gilman wrote:
> >to follow up on what Graham Klyne said:
> >
> >> I am basically promoting default to SI unless explicitly
> >> overridden (at per-use or per-feature level or both is under
> >> debate).
> >
> >You may be happier with the results if a bare number for a
> >feature whose registration (or protocol binding) did not assert a
> >default unit is an error.
> >
> >There are two qualitatively different error modes here.  In
> >one the setter of the attribute does not understand the SI default
> >and the reader of the attribute receives erroneous information.
> >In the other there is a processing error at some point, but not
> >the undetected exchange of bad information.
> >
> >This is a judgement call.
> >
> >To use SI units as the default, the registration would have to
> >declare the dimension of the quantity anyway.  (length vs. mass
> >vs....).  People are more likely to give you a unit than a
> >dimension, in practice.
> >
> >Al
> >
> >
> ------------
> Graham Klyne
>-- End of excerpt from Graham Klyne