[Groff] quasi device independence of intermediate output

[Bernd Warken]

On Sat, Nov 17, 2001 at 10:39:28PM -0000, Ted Harding wrote:
> The best device ("-T<device>") to use with xditview
> is ps: ... | troff -Tps .... | xditview
> 
> You can also (and probably better) invoke xditview
> using groff's "-X" option:
> 
>    groff -Tps -X ....
> 
> Note: Paradoxically, TeX's output format (DVI) is
> also device-independent; it is converted to a particular
> device rendition by a suitable program (e.g. dvi2ps
> for conversion to PostScript). However, DVI is also
> a device as far as gtroff is concerned ("-Tdvi"),
> troff's output being converted to .dvi by grodvi.
> 
Classical (di)troff had a very strange kind of device independence.
troff output for some device could be handled by an arbitrary other
postprocessor device.  This seems to have been done by a scaling
of all arguments in basic units by the ratio of the resolution of
the inflicted devices.  Apart from rounding errors, this seemed to
have worked.

Basically, groff does not support this.  It might be not too hard to
implement this.  The right place for this is the output parser.  The
name of the postprocessor's native device would be a global variable
that is used by the parser to determine a resolution quotient.  That
would be applied to the suitable commands within the browser.  The
device name in the intermediate output document could then be discarded
and the postprocessor device name be used instead.  This could enable
even the processing of very old devices, e.g. "post", without knowing
their other parameters or fonts.

Does this concept make sense?

Bernd Warken