Re: Re: texfont and type-tmf.dat

Re: Re: texfont and type-tmf.dat

[George N. White III]

On Tue, 14 Oct 2003, Nigel King wrote:

> > Why do you whish to reinstall all the supplied fonts? Or do you just
> > want to learn how to use texfont?
> I simply wish to use the fonts as most others do (I think) and I am unable
> to because of this bug. The normal context fonts do not exist on Gerben's
> version. Before Gerben's reorganisation they were made by texfont, this now
> does not work because of what I believe to be a simple (I do not understand
> Perl otherwise I would have a go at fixing it) bug in texfont. Maybe I am
> wrong but I am beginning to find this extremely frustrating (7 months). Once
> I realised that somebody was working on texfont 1.8 I had hoped for a fix.

On all the systems I use, (linux, SGI Irix, Win32 with texLive) you have
to run texfont to generate tfm's, vf's and map files for fonts other than
CM.

The last time I tried to run texfont (from the final TeX Live 2003 inst
iso), it tried to use a lower-cased version of the install path, e.g.,
/opt/texline instead of /opt/texLive.

> If no developer/Perl expert has Gerbens version of tetex perhaps I could
> help debug if a version was supplied to me that printed out the full path
> that texfont was looking for files in. The --fontroot option or kpsewhich
> lookup appears to be at fault but I have no debugging facilities.
>
> Any suggestions (other than giving up) gratefully received.

You will have to learn a tiny bit of perl (which is easier than
learning TeX).

-- 
George N. White III  <aa056@chebucto.ns.ca>

Re: texfont and type-tmf.dat

[Patrick Gundlach]

Hi,

> The last time I tried to run texfont (from the final TeX Live 2003 inst
> iso), it tried to use a lower-cased version of the install path, e.g.,
> /opt/texline instead of /opt/texLive.

which should be unnecessary on MacOS X (when using hfs as a filesystem)

Patrick
-- 
You are your own rainbow!

Re: Re: texfont and type-tmf.dat

[Nigel King]

>> I simply wish to use the fonts as most others do (I think) and I am unable
>> to because of this bug.
> 
> Nigel,
> 
> did you try 
> 
> \usetypescriptfile[adobekb]
> \setupencoding [default=8r]
> \usetypescript[adobekb][\defaultencoding]
> \usetypescript[palatino][\defaultencoding]
> \setupbodyfont[palatino]
> \starttext
> This is crazy stuff.
> \stoptext
Yes this is the only thing I can do currently, but I thought I had more
control previously.
> 
> ??
> 
>> The normal context fonts do not exist on Gerben's version.
> 
> What are normal ConTeXt fonts?
Say ec-urw-times and the typescripts etc.
> 
> 
>> Maybe I am wrong but I am beginning to find this extremely
>> frustrating (7 months).
> 
> I know how frustrating fonts in TeX can be :-(
> 

-- 
Nigel

Re: Re: texfont and type-tmf.dat

[Nigel King]

> Why do you whish to reinstall all the supplied fonts? Or do you just
> want to learn how to use texfont?
I simply wish to use the fonts as most others do (I think) and I am unable
to because of this bug. The normal context fonts do not exist on Gerben's
version. Before Gerben's reorganisation they were made by texfont, this now
does not work because of what I believe to be a simple (I do not understand
Perl otherwise I would have a go at fixing it) bug in texfont. Maybe I am
wrong but I am beginning to find this extremely frustrating (7 months). Once
I realised that somebody was working on texfont 1.8 I had hoped for a fix.

If no developer/Perl expert has Gerbens version of tetex perhaps I could
help debug if a version was supplied to me that printed out the full path
that texfont was looking for files in. The --fontroot option or kpsewhich
lookup appears to be at fault but I have no debugging facilities.

Any suggestions (other than giving up) gratefully received.

TIA
-- 
Nigel

Re: Re: texfont and type-tmf.dat

[George White]

On Thu, 9 Oct 2003, Patrick Gundlach wrote:

> [...]
> According to to the statements from Walter Schmidt, a TeX font expert
> (perhaps I should say *the* TeX font expert?) in
> http://tug.daimi.au.dk/archives/tex-fonts/msg01328.html
> 
> \quote{% ....
> Note, however, that embedding of URW's fonts, while using the 
> (PSNFSS) Adobe Base35 metrics, will _not_ lead to any bugs!  
> The character metrics are matching!  Differences in the 
> "character bounding boxes" are irrelevant for the advance widths!  
> The only drawback is, that you cannot access those glyphs that 
> are in the URW fonts, but not in the Adobe fonts.  Indeed, this 
> could be overcome by providing particular metrics and VFs for 
> the URW fonts -- see below. }

Hans has demonstated that even the Adobe fonts don't have the same
metrics.  It should also be noted that in practice, if you don't embed
fonts, you will often get font substitutions in the PS rasterizer (e.g.,
ghostscript defaults will use URW fonts where the file requests a Base35
font, current acrobat reader will use Arial where the file requests
Helvetica, some printers with clone interpreters (many recent HP models)
use "clone" fonts. 

There are several versions of the URW fonts in use now: two ghostscript
versions, and a number of versions with additional glyphs distributed
with linux (and I am told that the software used to create the recent
versions may have tampered with the metrics for glyphs that were not
changed).

If you embed the URW fonts using the original URW names it is clear which
fonts are to be used.  This discourages people from "optimizing" your
files by stripping out the fonts.  For archival EPS figures it makes sense
to go further and replace fonts with outline paths.  In this way the
figures should remain useful even after the fonts are no longer supported
by the available rasterizers. 

--
George White <aa056@chebucto.ns.ca> <gnw3@acm.org>
189 Parklea Dr., Head of St. Margarets Bay, Nova Scotia  B3Z 2G6

Re: Re: texfont and type-tmf.dat

[Hans Hagen]

At 11:00 09/10/2003, you wrote:
>Hi,
>
>
>Hans Hagen <pragma@wxs.nl> writes:
>
> >>So what is wrong with adobekb.tex? It is now included in ConTeXt, so
> >>with \usetypescript[adobekb][someencoding] your psnfss fonts should
> >>work fine.
> >
> > i can make a default for that but only when i can be absolutely sire
> > that the metrics are there that i want there to be -)
>
>According to to the statements from Walter Schmidt, a TeX font expert
>(perhaps I should say *the* TeX font expert?) in
>http://tug.daimi.au.dk/archives/tex-fonts/msg01328.html
>
>
>\quote{% ....
>Note, however, that embedding of URW's fonts, while using the
>(PSNFSS) Adobe Base35 metrics, will _not_ lead to any bugs!
>The character metrics are matching!  Differences in the
>"character bounding boxes" are irrelevant for the advance widths!
>The only drawback is, that you cannot access those glyphs that
>are in the URW fonts, but not in the Adobe fonts.  Indeed, this
>could be overcome by providing particular metrics and VFs for
>the URW fonts -- see below. }

hm, there are more font experts (nelson b, boguslaw j, adam t, tom k, some 
people on this list as well, to mention a few), here is what Nelson Beebe 
says about this topic in a different thread ..


 > > Walter confirms what I have assumed: we can use the "Adobe" metrics from
 > > the PSNFSS bundle with the URW fonts.

I have serious reservations about this.

While as far as I know, there has been only one release of URW fonts,
and thus only one pair of (.pf[ab],.afm) files for each typeface, with
Adobe fonts, over the last 19 years, there have been silent changes
made to at least AFM files for many fonts, including Times-Roman.

This makes me suspect that the base-14 or base-35 fonts embedded in
tens of millions of laser printers with Adobe PostScript
implementations may in fact not be identical, even though they share
common font names.

Of course, the changes are usually pretty small, and few people would
ever notice.  However, precise character positioning demands knowledge
of metrics, and if a TeX job uses metrics which differ from those
embedded with a font in a printer, and uses the resident fonts, rather
than downloading them, then output will certainly not be what TeX (and
the user) intended.  That is one reason why I've never been entirely
happy with fontless PostScript and PDF files, and why I was
exceedingly unhappy with the change in Adobe Illustrator last year
that completely ignores embedded fonts, and uses only installed fonts.

The program MIME-attached below can be used to compare AFM files, and
I have just done so with the
texlive7/texmf/fonts/afm/{adobe,urw}/times/*.afm files.

The first thing to note is that the URW fonts contain many more
glyphs: 316 for Times-Roman compared to Adobe's 228.  A TeX file that
used any of the additional URW glyphs would print incorrectly with
Adobe's Times-Roman.

The second thing is that the bounding boxes can be a bit different,
and sometimes very different, even when the widths are identical:

         % awk -f afmdiff.awk /tmp/afm/adobe/times/ptmr8a.afm 
/tmp/afm/urw/times/utmr8a.afm
         Comparison of AFM metrics in files: 
/tmp/afm/adobe/times/ptmr8a.afm /tmp/afm/urw/times/utmr8a.afm
         Font names: Times-Roman NimbusRomNo9L-Regu
         ...
         WX width differences:

         Bounding box width differences:
         dagger            -1       dieresis          -1       dotaccent 
      -1
         exclam             1       exclamdown         1       Idieresis 
      -1
         idieresis         39       three             -1 
threesuperior     -1

         Bounding box height differences:
         Adieresis          1       adieresis          1       aring 
     -10
         Aring            -17       asciitilde         6       dieresis 
       1
         dotaccent          1       Edieresis          1       edieresis 
       1
         exclamdown        -2       greater           -4       Idieresis 
       1
         idieresis          1       less              -4       Odieresis 
       1
         odieresis          1       plusminus        -62       q 
      -1
         questiondown      -2       s                  1       Udieresis 
       1
         udieresis          1       Ydieresis          1       ydieresis 
       1

TeX uses more than just the bare width, so I suspect that we can
readily demonstrate different typesetting with these two
purportedly-compatible Times-Roman fonts from URL and Adobe.  As a
simple experiment, I created two DVI files with "tex testfont", like
this:

         % tex testfont
         This is TeX, Version 3.1415 (C version 6.1)
         (/usr/local/lib/tex/inputs/testfont.tex
         Name of the font to test = ptmr8r
         Now type a test command (\help for help):)
         *\table

         *\bye
         [1]
         Output written on testfont.dvi (1 page, 10632 bytes).
         Transcript written on testfont.log.
         % mv testfont.dvi testfont-ptmr8r.dvi

and similarly for utmr8r.  I then ran dv2dt on both, and compared the
output:

         % dv2dt < testfont-utmr8r.dvi > testfont-utmr8r.dt
         % dv2dt < testfont-ptmr8r.dvi > testfont-ptmr8r.dt
         % diff testfont-utmr8r.dt testfont-ptmr8r.dt
         19c19
         < (utmr8r)
         ---
         > (ptmr8r)
         33,34c33,34
         < d3 1518460
         < d3 1911676
         ---
         > d3 1512892
         > d3 1906108
         191c191
         < fd1 50 24364160751 655360 655360 0 6 'utmr8r'
         ---
         > fd1 50 4767720433 655360 655360 0 6 'ptmr8r'
         291c291
         < d3 721220
         ---
         > d3 715652
         306c306
         < sr 1016132 26214
         ---
         > sr 1010564 26214
         ... many more ...


Clearly, the DVI files differ somewhat.

I did similar AFM comparison experiments with the 3 other times fonts
in TeXLive 7:

         % awk -f afmdiff.awk /tmp/afm/adobe/times/ptmri8a.afm 
/tmp/afm/urw/times/utmri8a.afm
         % awk -f afmdiff.awk /tmp/afm/adobe/times/ptmbi8a.afm 
/tmp/afm/urw/times/utmbi8a.afm
         % awk -f afmdiff.awk /tmp/afm/adobe/times/ptmb8a.afm 
/tmp/afm/urw/times/utmb8a.afm

All had identical character widths, but many differences in bounding
boxes.

Thus, I believe that it would be wrong to claim that the URW fonts are
true drop-in replacements for the Adobe fonts, and of course there
will be minor shape differences as well, some of which may be visible
to sharp-eyed readers, at least in heading-size characters.

As an additional experiment, I ran the locate command on our large
Unix installation, and found 64 instances of Times-Roman.afm.  MD5
checksums of these files showed that there are only 7 different ones,
so I ran the afmdiff.awk program on those 7.  While there were no
differences reported in character widths, there were many differences
in bounding boxes and even in glyph counts (210, 228, and 315,
depending on the file):

The copyright statements look like this:

Comment Copyright (c) 1984 Adobe Systems Incorporated.  All Rights Reserved.
Comment Copyright (c) 1984 Adobe Systems Incorporated. All Rights Reserved.
Comment Copyright (c) 1985, 1987, 1989 Adobe Systems Incorporated.  All 
rights reserved.
Comment Copyright (c) 1985, 1987, 1989, 1990 Adobe Systems 
Incorporated.  All Rights Reserved.
Comment Copyright (c) 1985, 1987, 1989, 1990, 1993, 1997 Adobe Systems 
Incorporated.  All Rights Reserved.
Notice Copyright (c) 1985, 1987, 1989 Adobe Systems Incorporated.  All 
rights reserved.Times is a trademark of Linotype AG and/or its subsidiaries.
Notice Copyright (c) 1985, 1987, 1989, 1990 Adobe Systems 
Incorporated.  All Rights Reserved.Times is a trademark of Linotype AG 
and/or its subsidiaries.
Notice Copyright (c) 1985, 1987, 1989, 1990, 1993, 1997 Adobe Systems 
Incorporated.  All Rights Reserved.Times is a trademark of Linotype-Hell AG 
and/or its subsidiaries.

Re: Re: texfont and type-tmf.dat

[Hans Hagen]

At 20:13 08/10/2003, you wrote:
>Hi Nigel,
>
> >> What are you trying to achieve?
>
> > I use Gerbens distribution of tetex and friends on Mac OSX. In Feb/Mar the
> > distribution was altered which broke texfont. I had to resort to using
> > adobekb.tex which works very well. I was hoping however at some point to be
> > able to use the fonts in a more standard way with everybody else.
>
>So what is wrong with adobekb.tex? It is now included in ConTeXt, so
>with \usetypescript[adobekb][someencoding] your psnfss fonts should
>work fine.

i can make a default for that but only when i can be absolutely sire that 
the metrics are there that i want there to be -)

Hans  

Re: Re: texfont and type-tmf.dat

[Nigel King]

Dear Patrick,
> What are you trying to achieve?
I use Gerbens distribution of tetex and friends on Mac OSX. In Feb/Mar the
distribution was altered which broke texfont. I had to resort to using
adobekb.tex which works very well. I was hoping however at some point to be
able to use the fonts in a more standard way with everybody else.

-- 
Nigel