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

Re: What's the relationship between vfs and tfms?

>The only case where this can get us in trouble is if some user is
>trying to achieve some effect by changing the font associated with
>a particular font name; that is, he or she wants his or her `cmr10'
>to be rendered in outline form, so he or she creates a cmr10.vf file
>that does the appropriate redirection.  If this is intended to be
>supported functionality, then yes, the way dvips does things needs
>to be reconsidered . . .

Related to this, have you come across this idea from Simon Cao, which
Andrew Treverrow forwarded to the oztex-info mailing list:

From: akt@netspace.net.au (Andrew Trevorrow)


>From Simon Cao <simonmc@ebox.tninet.se>:

Let me first give my opinions on the type1cm.sty and VF solutions
to the T1/MF compatibility problem. I'm afraid I can not agree
with most part of what George Gratzer addressed in his mail
dealing with AMS PS font headaches. I should say that I have been
using David Carlisle's type1cm package occasionally since its
appearance on CTAN. In my opinion, type1cm.sty is only a primitive
and limited solution to the T1/MF compatibility problem.
The reasons are:
1. type1cm.sty is based upon LaTeX2e at a higher level. Users of
        plain TeX, LaTeX209 and other macro packages cannot benefit
        from it. It may not be an essential and integrated solution to
        the problem.
2. DVI files made with MFs of design sizes that are not presented
        in T1 fonts still can not be exchanged with a T1 TeX system. I
        can't see how this can be fixed by the type1cm and other
        methods except for VFs. We recall that DVI files are supposed
        to be device independent!
3. Worst of all, when a LaTeX2e file including non-T1-design-size
        is transformed to a type1cm file, character spacing, line
        break, page break and graphics position will change, due to
        change of font metrics. We can surely view and print the LaTeX
        file by this method. But do we preserve the layout of the
        original file and the design idea of the original author? This
        is the essential point of our beloved TeX system and is
        strongly addressed by the author of TeX, Donald Knuth!
4. By another scan through the readme file included with type1cm
        package, the size rendering mechanism is different from that
        of the AMS amspsfonts option. Of cause this can be modified.
        However, David has no intention to provide his package as a
        solution to the T1/MF compatibility problem, merely a
        convenient way for T1 users to adjust their systems for better
        viewing using T1 fonts and the freedom to scale the T1 fonts
        to any size. The latter yet introduced further incompatibility
        as David himself realized the issue. In short, type1cm package
        is intended for personal choice only.

By using VFs, these problems do NOT exist and we can use the
original tfm files produced by the AMS MFs.  Most importantly the
VF mapping mechanism is native in any modem TeX systems and is
clearly a choice of Prof. Knuth himself for the flexibility and a
great feature of TeX. What do we expect, in the particular case of
VFs for the AMS/PS missing sizes, is that tuning of individual
characters in the VPL files has to be done using the powerful
parameter settings provided by VF, in order to represent the
original MF design as much is possible.  It is possible, however
requires time and energy. It can be done once for all. The worst
thing one can get is a deviation of a scaled fonts to their MF
counterparts, but font metrics, line break, page break, layout
will be preserved! This is exactly modem layout viewing
applications are doing, such as the Adobe Acrobat Reader and the
SuperATM font substitution mechanism. TeX files and DVI files
based on different systems can be exchanged without any problem.
And one can even switch between T1/MF setups for printout and for
producing online document (for example PS->PDF). I personally
prefer print out using MF bitmaps, and viewing or producing PDF
using T1 fonts.  Interesting enough, the plain tex file
"amsfndoc.tex" included in the AMS/PS distribution can not be
texed or viewed using only the T1 font setup without VFs! Do you
believe this?

These are only my personal opinions of this current issue. If you
find it worth of these words, feel free to forward this message to
the oztex-info related discussion.

Now lets return to the files I included in the attachment.
1. psfonts.CMPS file is the DVIPS mapfile corrected some misprint
        and added missing entries upon the original one in the OzTeX
        distribution. I mentioned to you in the last message.
2. VFs and corresponding VPL files for all missing design sizes in
        the AMS/PS distribution.
3. test files for all the AMS/PS VF fonts along with their true
        size T1 fonts. I used them to test if my VFs work. They were
        also used to produce DVI files for viewing and postscript
        files for printing on T1 only and MF only systems for visual

The included VFs and VPL files are simply mapped to the available
T1 fonts by scaling, according to the standard amspsfonts
rendering definitions. I tested them using the test files I
included. I found out that the printouts from VF and MF setups
have no appealing visual differences for my eye-sight, except for
darker (fatter strokes) characters using T1 fonts which was
expected even for larger design sizes. Another point is that the
AMS small-size fonts are mainly intended for math, so that small
difference in their character shapes does not make too much sense
at such small sizes. Perhaps, one can do more with the VPL files,
but I really have no such energy. I included the test files so you
can compare by yourself and make your own judgment.

Here is a small story about BSR's Textures related to my VFs. Last
time I told you I cannot make small-size VFs for WNCY* series AMS
fonts. First. I thought the scaling of 10pt fonts may not work
with WNCY's, since WNCY's have only 10pt T1 fonts while MFs are
designed down to 5pt. Then, it took me some time to figure out
that the VF metric files provided on their ftp site are incorrect!
I verified this using both tetex and OzTeX3.1. I also wondered why
BSR do not provide the VFs for the WNCY's. To see what is wrong, I
then used Textures Reader 1.8.2 and Textures 1.6.3 (our dept. has
only this old version on an old Mac) to view and print the test
DVI files. Textures Reader produced exactly the same result as
that of OzTeX and tetex systems. However, Textures 1.6.3 gives
different result. For AMS fonts other than the WNCY's, rendering
is always possible to their nearest or next brothers. So one like
me with less sensitive eyes(?) may not notice that.  But the 5pt
WNCY's have to be scaled from 10pt, which produced a printout of
2.5pt instead of 5pt with their VF scaling scheme. They probably
also noticed this fact and either don't how to fix the VPLs or
have not yet fixed the VF bug in the Textures 1.6.3 at that time.
Thanks god they fixed it in Textures 1.8.x (perhaps 1.7.x?).
However, they do not change the bad VF metrics provided on their
ftp site. These poor VPs have been fixed in the my files send to
you. Now all VFs for all missing design sizes in AMS/PS are