[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

dvi to PS-type1: assumptions, strategies

dvi to PS-type1: assumptions, strategies

Discussions of drift for dvi-to-PS drivers seem to assume the
following about Adobe type 1 fonts.

The PS interpreter always lays down the same bitmap for a given
character, and a string of characters lays down the string of
corresponding bitmaps.  Further the pixel width of the bitmap for an
individual character is the naive rounding of the width in the .tfm
which will agree with widths in the PS font.

However this assertion is procedural and not declarative so I doubt
Adobe has or will ever make such a statement. I even have some
doubts that it is true on existing equipment. The interpreter is
free to play its own drift limitation game --- and that would make
nonsense of the traditional game played by TeX drivers. (A PS based
machine (NEXT) should let one see the answer in one case.)

    There are fortunately drift control strategies that should be
effective independent of resolution, and of drift control for
character strings (conjecturally) applied by the PS interpreter.

(a) There is a "incessant absolute drift zeroing" policy in which all
drift management is  skipped by the TeX driver and every object is
assigned its nearest *absolute* pixel position on the page. Andy
Trevorrow has repeatedly mentioned this approach. (What is the
characteristic PS syntax here?) Is this policy too costly timewise?
Is this *exactly* what OzTeX does or is Andy's approach more
complex? It would be helpful to know what it achieves.

(b) "cued absolute drift zeroing". There is another approach where an
item is nailed to its absolute position only on a number of specific
cues (I am seeking to establish a list of such --- big horizontal
skip is one; drift exceeding a maxdrift value is out if as i am
suggesting we do not know what the drift will be.)

      larry siebenmann