> currently I haven't seen any arguments why mltex
> primitives are technically superior to vf
There are two big reasons:
1) default use of ML TeX produces standard Knuthian output, namely DVI
files (a Fuchs-Knuth standard) appealing to standard CM fonts (Knuth).
There is no electronic document format that is overall better re
compactness, usability, and long term stability. The
use of VFs leads to DVIs that involve AE fonts that may change and are
not widely installed. Conversion AE ==> CM of the DVI using
the utility dvicopy yields a CM based DVI that is many times bigger than
the DVI from MLTeX and also slower...
2) A sophomore TeX user is able to tweak the accent definitions in MLTeX
to achieve useful typographic results for an individual job. The similar
change using VFs requires a font gourou and possibly several jobs to
amortize the the time investment.
> that [VF > MLTeX] presumably was Knuth's opinion at the
> time, mltex predates vf, doesn't it?
There are no grounds to presume. Knuth went to great pains to assert
Fuchs' priority (over Adobe Inc, J. Warnock?) in the introduction of the
VF concept. The clear priority of Ferguson on the MLTeX idea is rather
(*I presume*;=) the reason for its exclusion from TeX 3.
For the future, I hope that (in addition to VFs) the MLTeX feature
will be in NTS. I really would want a more flexible MLTeX feature where
for example accent definitions can be successfully changed inside a
paragraph --- say in switching between French and German.
PS. I have not noticed the slightest typographical difference between
the accent typography of AE fonts and that produced by Knuth's accent
control sequences (or default MLTeX), but I have not examined all
details by any means! AE author Lars Engebretsen, <[log in to unmask]>
might be receptive to criticism.