Media-features: proposal for document size

[Graham Klyne <GK@xxxxxxxxxxxxxx>]

In an effort to try and get the document/image size debate resolved as
quickly as possible, I include below a proposal to replace 'pix-x' and
'pix-y' in the Media-features draft.  It is based on the idea of using
pixels-per-resolution as the measure of document size.

Comments, please!

#g
--


2.1 Document Image Size

   - Media Feature tag name(s): 

     size-x
     size-y
    
   - ASN.1 identifier associated with this feature tag:

     ***New assignments by IANA***

   - Summary of the media features indicated by this feature tag:

     These features indicate the size of a displayed, printed or otherwise
     rendered document image;  they indicate horizontal (size-x) and
     vertical (size-y) dimensions.

     The unit of measure is pixels/resolution;  thus, if the declared
     resolution of the rendered image is in pixels/inch, then the image
     dimensions indicated by these features are in inches.  In the absence
     of a declared resolution a value or 1 is assumed, thus giving the
     image dimensions in pixels.

     If different resolutions apply in the x- and y- dimensions, then the
     horizontal resolution is used in interpreting the size-x value, and
     vertical resolution is used in interpreting the size-y value.

   - Values appropriate for use with this feature tag:

     Rational, >0

   - The feature tag is intended primarily for use in the following
     applications, protocols, services, or negotiation mechanisms:

     Display and print applications where different media choices will
     be made depending on the size of the recipient device. For
     example, a web application for use on a 240x480 display might use
     different HTML pages than one intended for use on a 1024x768
     display, or on a print device with an 8 inch by 10 inch imageable
     area.



NOTE:

In a dimension-less (or unit-less) environment, resolution would not be
included in a capability description, so any size- measurements would be
interpreted as pixels.  In a dimensioned environment, specifying a
resolution allows the document size to be interpreted as a physical dimension.

It is possible to mix both modes in a single capability statement;  e.g. a
web client with a VGA screen and a 300dpi printer might have the following
capabilities:

  (| (& (size-x<=640) (size-y<=480) )
     (& (size-x<=8) (size-y<=10) (dpi=300) ) )

To describe a drawing that is to be rendered as as a 10x15 inch image with
a resolution of at least 100dpi, the following might be used:

  (& (size-x=10) (size-y=15) (dpi>=100) )

Note that, in this case, it is the resolution *units* that are important,
not the actual resolution value as that would cancel out when the size
values are calculated.  In the above example, if the actual resolution
were, say, 123dpi then the document width would be 10x123 pixels, which at
123dpi is exactly 10 inches.


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