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

free slots




 *** free slots ***

Recently J"org commented:

 > Atomicness can be used to create a family of standard encodings
 > a little similar to the ISO 8859 family of standards. However,
 > such a family should have as few members as possible. I don't
 > support the idea of ,,free slots'' which can be arbitrarily
 > filled in.

Perhaps J"org could explain the analogy with the ISO 8859 family of
standards.

I do not myself support the idea of ,,free slots which can be arbitrarily
filled in''.  What I do support is the idea of free slots that will
possibly filled in but only by some higher level of the same standard.

The alternative to this gradualism is to immediately fill extra space
with trash, or to speak more mildly, with useful material that really
belongs elsewhere. That is what Don Knuth did with CM; thus the basic
prose font cmr10 contains some upright greek capitals for mathematics. I
beg you consider my contention that if Knuth had allowed only clearly 
relevant characters into cmr10, and requested that no individual change
the fonts without renaming them, then SURELY the Cork norm would be
essentially compatible with the CM encodings.  In other words,
Knuth's "full at all costs" approach is responsible for much of the pain
that is attending transition to the font system now taking shape.
(It also prevented Europeans being decently served by TeX2
in the 1980's.)

    A font system that cannot conveniently be updated in an upwardly
compatible way will have a short lifespan --- or alternately suffocate
its offspring in a wretched old age.  It is not quite sure which will be
the case for CM. In any event, I sincerely hope that the same "full at
all costs" mistake is not made with the atomic encodings I am proposing.

    It is also worth discussing how the math font system will be updated.

    This reproach cannot be addressed to the Cork encoding since there
was too little space from the outset. Nor to J"org's African fc encoding
since it too ran out of space. In each case, one might say there was an
optimal way to do the impossible --- almost.

    In the case of the atomic fonts there seems to be enough space to do
a fairly thorough job; that is a joy but also a tempation to repeat
Knuth's mistake.

           Laurent Siebenmann