[tex-live] urw35 base fonts on CTAN and TL

[texlive at schoepfer.info]

On 2017-12-16 00:28, Reinhard Kotucha wrote:
> On 2017-12-12 at 17:09:45 +0100, texlive at schoepfer.info wrote:
> 
>  > On 19.09.2016 02:14, Reinhard Kotucha wrote:
>  > > On 2016-09-19 at 00:10:08 +0900, Norbert Preining wrote:
>  > >
>  > >   > Hi Staszek,
>  > >   >
>  > >   > > E.g. on January 2015 there was much better release, and from 
> July
>  > >   > > 2016 it seems that fonts are finally proper. Moreover the 
> whole
>  > >   >
>  > >   > AFAIR these "new" fonts changed the metrics compared to 
> previous
>  > >   > versions of some glyphs. Since these fonts are used as 
> replacement
>  > >   > for the Base35 fonts, they need to be metric wise equal. But 
> they
>  > >   > aren't.
>  > >   >
>  > >   > Then there is the addition of completely insufficient and 
> broken
>  > >   > cyrillic... but I see that they have at least added the 
> cyrillic
>  > >   > glyphs now for all fonts.
>  > >   >
>  > >   > I am really not sure what is the best way to proceed, and 
> maybe the
>  > >   > new fonts are again metric wise compatible ...
>  > >
>  > > Hi Staszek and Norbert,
>  > > the fonts do not only have to be metric compatible, they have to
>  > > provide exactly the same sets of glyphs as the original fonts 
> released
>  > > by URW and maintained by Walter Schmidt.
>  > >
>  > > The ghostscript fonts shipped with TeX Live are the same as those
>  > > which are part of the psnfss LaTeX package.  And psnfss provides 
> .tfm
>  > > files for exactly these fonts.  This is why I maintain these fonts 
> in
>  > > TeX Live at all.
>  > >
>  > > The main problem is that the fonts were extended but their 
> internal
>  > > names (the /FontName variable) were not changed.  The Type 1 font
>  > > specification (Adobe) clearly says that there shall never exist 
> two
>  > > different fonts with the same /FontName.  For a good reason!
>  > >
>  > > Some time ago someone told me that he created a PostScript graphic 
> but
>  > > a particular glyph didn't appear in the document created with 
> LaTeX
>  > > though he could see this glyph in a PS viewer.  It took me some 
> time
>  > > to find out what happened.  It turned out that he used a glyph 
> which
>  > > wasn't supported by psnfss, pdftex assumed that the font provided 
> by
>  > > ghostscript and that in the texmf tree are identical and 
> substituted
>  > > the font.  Such kind of problems are quite difficult to track 
> down.
>  > >
>  > > I'll look into the new fonts anyway.  But I don't think that we 
> can
>  > > use them because they are not compatible with what we have in TeX 
> now.
>  > >
>  > > If they turn out to be useful, the only way to make them available 
> to
>  > > the TeX world is to rename all these fonts (/FontName), create TeX
>  > > support files for them, and create two packages, one for TeX and 
> one
>  > > for ghostscript.  I currently have no idea how to make such fonts
>  > > accessible to ghostscript on all platforms.
>  > >
>  > > Please note that the problem I described above occured on Unix.
>  > > Windows users are in advantage because both, TeX and ghostscript 
> are
>  > > using exactly the same fonts.  On Unix there is currently no way 
> to
>  > > avoid such problems because TeX Live can't provide an adapted
>  > > ghostscript installation for all supported platforms.
>  >
>  > If this works on Windows...
>  > Is it correct, that if a linux distribution provides another/newest
>  > version of urw-core35 type1 fonts and ghostscript, there is no 
> Problem
>  > with metrics or "same sets of glyphs" when texlive/psnfss uses the 
> same
>  > fonts by changing the corresponding map-files?
>  > I assume generating font description(fd) files and virtual fonts(vf)
>  > would also be necessary, or would there be much more to be done from 
> a
>  > linux distribution point of view?
> 
> Hi Johannes,
> I fear that there is no satisfying solution.

Thank you very much for the detailed explanation!

> We currently have three
> different fonts with the same internal variable /FontName.  Each
> supports a different set of glyphs.
> 
>  1. TeX Live provides and supports the fonts gratefully donated by
>     URW.  None of the fonts were modified except NimbusSanL-ReguItal
>     (uhvro8a.pfb) due to a severe bug in /germandbls, see
> 
>       http://tug.org/~kotucha/germandbls.pdf
> 
>     The original font is still available (uhvro8a-105.pfb).
> 
>  2. Ghostscript has subsets of these URW fonts built into the binary.
>     These fonts are used if there is no file called "Fontmap" in
>     Ghostscript's search path (GS_LIB).  Many glyphs were removed from
>     the URW fonts but an additional glyph was added to the symbol
>     font.
>     According to Chris Liddell, a Ghostscript developer, the sole
>     reason to provide any fonts at all is to have a replacement for
>     the Adobe fonts built into PostScript printers.  Thus glyphs not
>     supported by Adobe were removed from the URW fonts.

With fontforge, i did a short compare of base35 font Times-Roman from 
ctan with the newest fonts-core35 release from 
https://github.com/ArtifexSoftware/urw-base35-fonts, which seem to be 
the fonts builtin in the newest ghostscript 9.22, by
sfddiff base35/pfb/utmr8a.pfb fonts/NimbusRoman-Regular.t1
There are a lot of differences and new glyphs in the newer font, but it 
seems the only glyph missing in the newer font is "commaaccent".

>  3. Ghostscript offers an external font package.  It comes with its
>     own Fontmap file and thus overrides the built-in fonts.  These
>     fonts come with additional (Cyrillic) glyphs.

I'm confused, you mean e.g this
https://packages.debian.org/stretch/gsfonts ?

> The problem is that all three fonts are incompatible but have the same
> /FontName.

[...]

> Nothing can be done in TeX Live in order to solve this problem.  Every
> font can be supported if the necessary files are available but there
> is absolutely no way to support different fonts with the same name.

If it's true that the github-artifex base35 fonts only misses one glyph 
per font,
Would it be possible to make the texlive psnfss package work with this 
fonts?
Just to know if i got it right: If it can be done in texlive, would this 
base35 problem be solved, if all distributions would "reset", and start 
over to only use the newest base35 realease?

> Let me know if you are interested in a font package providing the
> original URW fonts for use with Ghostscript.  During the next three
> weeks I'm on leave and thus have more time as usual.

Thank you very much for the offer, but i think i can do this on my own 
in case.

Johannes