bug in fontinst 1.801: duplicated kernings (was: ae fonts Q)

Bernd Raichle raichle@Informatik.Uni-Stuttgart.DE
Wed, 30 Sep 1998 11:00:53 +0200 (MET DST)


On , 30 September 1998 12:36:18 +0400,
Vladimir Volovich <vvv@vvv.vsu.ru> writes:
[...]
 > as far as i understand, implementing different fontinst behavior (for
 > original TeX and for e-TeX) is not needed: it is sufficient (and more
 > simple/preferable) to implement just the discussed extension to
 > fontinst.

Right!

 >           The original TeX anyway will not be able to use these
 > extended kerning features, will it? So, e.g. if some _font_ contains
 > kern between two ligatures, and this font is loaded by "original" TeX,
 > then TeX will behave exactly like it did before (i.e. it will be able
 > to create ligatures and will not be able to insert a kern between
 > them). But e-TeX being feeded with the _same_ font, will benefit of
 > this? Or am i wrong?

You are right.  For the original TeX only one of several lig/kern
instructions is used, all others are ignored.  TeX will simply take
the first instruction with matching glyph pairs.

For the (e-)TeX extension it will makes a difference because it
distinguishes between lig and kern instructions, therefore it can make
use of a kern even if there's a ligature instruction first (which can
be suppressed, for example, by a new primitive \suppresslig).  The
exact behaviour of this TeX extension in detail is still under
discussion (e.g. how to switch it on/off based on the current font).

Best wishes,
  -bernd