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

archiving format contest: .dvi versus .tex and .ps (2)


J"org's conclusion was

 > To my experience, only TeX source code in seven bit
 > ASCII code is really portable. Second best is a dvips
 > postscript file for 300dpi.

and his second given reason was

 > The second, os-independent problem is, that fonts are site
 > specific. You cannot really trust that the recipient has the
 > same fonts as you. Recently (about 6 month ago) I retrieved
 > a paper in dvi format, which still includes the looooong
 > outdated msxm and msym fonts. The result was a printout with
 > missing characters. I asked for the TeX source code or a
 > dvips ps file, but did get no reply from the authors.
 > 
 > There are other fonts which are withdrawn now, but may
 > survive in dvi-files forever, like the 14 point versions of
 > some AMS fonts.
 > 

Again I am unconvinced.

The msxm and msym fonts are *recently* outdated.  And the
distributions of msam, msbm replacing them may still be imperfect
according to some reports.  But that only *aggravates* the problem,
inasmuch as conservative users may still have some excuses to use
them.

The CM face should be firmly committed to durable standards and/or
upward compatibility, on the 100 year timescale mentioned as a goal
by Knuth.

Since msam, msbm were not based on 1982 metafont, they are
non-standard stragglers.  They should be excluded from a
scientific .dvi archive.

But what should an archive manager do if a preexisting .dvi using
them is submitted to a future archive and no useable .tex source is
available?

Desirable answer:  Use dvicopy and virtual fonts called msxm10,
msym10 etc etc to replace msxm and msym characters by corresponding
msam, msbm characters.  I believe this would work reasonable well
in practice.  The archive customer would usually notice nothing and
believe msam, msbm had been used from the outset.

Warning to creators of the the new math fonts:  Be absolutely
certain that msam, msbm can be replaced in .dvi archives by the the
new math fonts, using the above vf method.  The only way I know to
be certain is to build and check the virtual fonts providing the
msam, msbm emulation along with the new fonts!

The systematic use of vf expansion to eliminate nonstandard fonts
is not the whole solution.  The distribution of standard fonts
should be associated.  An archive for CM face fonts (.mf format)
should be somehow associated with the .dvi archive.  It should 
also be possible to have a server distribute .pk files generated on
the fly by metafont.

Archives based on .tex files probably have significant roles to
play. One would be to gradually impose upward compatibility on
successive versions of .tex support files by programmers who tend
to thumb their noses at the past.

                 Laurent Siebenmann