[tex-live] TeXlive infrastructure

Frank Küster frank at kuesterei.ch
Thu May 4 09:00:28 CEST 2006

karl at freefriends.org (Karl Berry) wrote:

> Hi Frank,
> Sorry, I'm pretty much at sea here as to exactly what problem you're
> reporting, and hence what solution you're proposing.

On the surface, the problem is just that the tpm files for the base 35
fonts (avantgar.tpm, ...) contain files that are not (no longer?) on
CTAN, and I have not found the license information and (fontinst)
sources for them.  Whether this will just resolve into "Ah, here they
are", and if it will require any moves (or even removals), this I don't
know yet, and therefore I'm not proposing any solution.  

One other thing is that CTAN and TeXlive are inconsistent with respect
to file placement, but wether this is a problem I cannot tell.

> - Ok, I missed the crucial word "metrics" at
>   http://www.ctan.org/tex-archive/help/Catalogue/entries/psnfss.html.
>   (I still fail to see any real reason why the type1's are separated
>    into a different package from everything else -- doesn't Walter
>    maintain it all?  But whatever.)

I think the reason is pretty much the same why any two related packages
on CTAN are in different directories:  They are of different origin,
have different authors, and license.  And I don't know whether anybody
really maintains the urw35 base fonts, since I doubt they are
developed.  The Palatino replacement that Ralf and Walter are preparing
will get a new name, while the existing fonts will be unchanged, AFAIK.

> - It doesn't seem to me that we should remove arbitrarily many font
>   files from TL just because they do not precisely correspond to
>   directories on CTAN (if that's what you're saying).

No, I must have been unclear.  First of all, removal would only be an
option if we can't get any license info for them, *and* if the fontinst
script to recreate them is not trivial.  But I guess it would be.  

>   That seems to me
>   like it will cause nothing but trouble.  It is just as likely that the
>   directories on CTAN should be more fully populated.

That's true, Walter is currently investigating some of the additional
files, and I'm sure he will comment on that soon.  But he has also
already said that he thinks it is unfortunate if the base35 packages in
texlive install additional TeX and LaTeX files that are neither in the
original urw-base35 distribution nor in psnfss.  I don't know his
reasons, but I guess one might be that he does not want people to expect
things from psnfss that it does not provide, and that he is not willing
to support.

So it might turn out that we keep these files were they are (and instead
update CTAN), or that we should consider to move them away from those
central packages to follow the wishes of the psnfss maintainer.  Let's
just wait for his answer; but in the meantime I'm searching for the
origin of those files.

> - I'm afraid I don't know what empty fonts/urw directory you're
>   referring to.  

I was referring to the directory on CTAN.  It seems I misunderstood you
when you said

| We couldn't make tpm's based on foundries (everything would end
| up in "public", etc.), and there is nothing else.  So each font, pretty
| much, ends up in its own tpm.

This I parsed as "we couldn't make a urw-base35.tpm, because..."  but I
see that this meaning doesn't make sense.  Personally, I don't propose
anything yet, but from Walter's mail I conclude that he would probably
prefer a mapping

 ftp://cam.ctan.org/tex-archive/fonts/urw/base35/ -> urw-base35.tpm

no matter whether the font files all end up in TEXMF/fonts/*/urw/ or in
subdirectories thereof, as it is currently.

>   In the TL repository, Master/fonts/tfm/urw has two
>   subdirectories: antiqua and grotesk.

Yes, but Master/texmf-dist/fonts/{type1,afm}/urw also contains
subdirectories for the base25 fonts.  The corresponding tfm files seem
to be in adobe - no idea why.

>  (As you probably know, those are
>   original free fonts from URW, before they donated the 35.)  There is a
>   sibling Master/fonts/tfm/urw35vf which has the base 35.

The last sentence is not the whole truth.  As I learned from Walter, the
files in urw35vf are *not* required to use the base35 fonts via psnfss -
the corresponding files are in fonts/tfm/adobe/.  The files in urw35vf
are the ones with "unknown origin".  

> - I already explained why it's simplest/best if the tpm's are created by
>   typeface, rather than by foundry.  (Because "public.tpm" doesn't sound
>   like a good thing.)  I suppose a special case for generating
>   urw35vf.tpm could be painfully programmed, but why?

I think here's a misunderstanding on your side.  On CTAN, we have in the
directory ftp://cam.ctan.org/tex-archive/fonts/urw/, without the zip

drwxr-xr-x   2 85       34           4096 Nov 17  2005 antiqua
drwxr-xr-x   2 85       34           4096 Mar 21 14:06 arial
drwxr-xr-x   5 85       34           4096 Nov 17  2005 base35
lrwxrwxrwx   1 85       34             32 Feb  3  2005 garamond -> ../../nonfree/fonts/urw/garamond
drwxr-xr-x   2 85       34           4096 Nov 17  2005 grotesq
lrwxrwxrwx   1 85       34             36 Feb 11  2005 lettergothic -> ../../nonfree/fonts/urw/lettergothic

and TeXLive has antiqua.tpm and grotesq tpm which install files in
fonts/*/urw/antiqua and fonts/*/urw/grotesq, respectively.  The obvious
parallel would be to have base35.tpm (or urw-base35.tpm, as in the
Catalogue) which installs into fonts/*/urw/base35.  I guess this would
be in fact straightforward to program, *however* we'd loose the urw35vf
fonts which are not on CTAN - and we should investigate first whether
they should be added to the base35 directory on CTAN, or somewhere

> - No real web interface was ever set up for the p4 repo, and I think the
>   p4 server is offline right now anyway.  However, all the log messages
>   (and changed file lists) are available in the current repo in
>   Build/cdbuild/logs/log.*, which is web-browsable at
>   http://tug.org/svn/texlive/trunk/Build/cdbuild/logs/

Ah, thanks, I'll look into this.

> I think I'm missing your basic goal -- what does it matter how big or
> little any given tpm is (within reason)?

The size isn't the point, correspondance with CTAN is it.

> And why single out this particular one for the base fonts?

As I pointed out above, it seems to me as if the base fonts are
currently singled out, in that the tpm structure doesn't correspond to
any CTAN structure at all.

> What are you really trying to accomplish?

Most importantly, just learn where these files come from and what their
license is.  If this also makes TeXlive structure clearer, and Walter
Schmidt happy, I'm glad about it.

Regards, Frank
Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX)

More information about the tex-live mailing list