size-x, -y proposal

[Graham Klyne <GK@xxxxxxxxxxxxxx>]

As they say:  post in haste, repent at leisure.

On reflection, I realize that my previously posted proposal for size-x,
size-y as an alternative to pix-x, pix-y is signifcantly flawed.

The specific problem I have noted is that a data resource described without
resolution cannot be effectively matched with a receiver that is described
with resolution.

Example:  consider a synthetic rasterized image which is dimension-free,
but constructed as 640x480 pixels, and a receiver with a printer with an
imagable aea of 10x8 inches and 300dpi resolution.  Using the proposal
previously posted, these are described by:

  (& (size-x=640) (size-y=480) )

and

  (& (size-x<=8) (size-y<=10) (dpi=300) )

Logically, the image is displayable on the printer, but the above feature
sets cannot be matched according to the feature set matching rules
currently specified.

Hence, I withdraw my previous proposal and shall seek another way of
accommodating the various issues this was intended to address.

#g

------------
Graham Klyne
(GK@xxxxxxx)