TL5 alpha3 (rep.2)

[Staszek Wawrykiewicz]

I. Basic installation on Linux:
-------------------------------
fmtutil: `pdfetex -ini -efmt=cont-de -progname=context cont-de.ini' failed.
install_link failed for /usr/TeX/bin/i386-linux/latex. File already exists.
install_link failed for /usr/TeX/bin/i386-linux/hugetex. File already exists.
install_link failed for /usr/TeX/bin/i386-linux/hugelatex. File already exists.
install_link failed for /usr/TeX/bin/i386-linux/texinfo. File already exists.

After installation, fmtutil.cnf in $TEXMFMAIN/web2c/ and even in 
$VARTEXMF/web2c/ has most of formats unblocked (but we have BASIC
installation!), so the first run of texconfig tries to build all that
formats. (???)
Again:
install_link failed for /usr/TeX/bin/i386-linux/latex. File already exists.

I've already proposed not to regenerate all formats when starting texconfig.
It is evident that we start texconfig to do something and it would be more
natural when closing it if some formats declared in fmtutil.cnf
are not ready.

lists/free/texlive1/texlive contains context ini files, but installed 
files are out of use when doing basic installation. This part should go to
lists/free/formats2/context

---------
Well, all was more clear when I started 
II. Recommended installation 
----------------------------

The first run of texconfig only simulates generating all formats; nothing 
new was added to $VARTEXMF/web2c/ or $TEXMFMAIN/web2c/ , even after
changes in fmtutil.cnf (all needed formats were already in $TEXMFMAIN/web2c/).
It is only pity that now $VARTEXMF/web2c/ does not contain a copy of
texmf.cnf. 

I always think what to do with VARTEXMF and HOMETEXMF issues...
As I wrote some times ago, the only way texconfig builds automagically 
a small local tree is setting VARTEXMF (setting HOMETEXMF or LOCALTEXMF
do nothing from the user's point of view; even more, in most cases, in 
standalone installations it is rather redundant). If VARTEXMF is not set
(most cases, because it is not stressed in the documentations or even
during install!), the user (or admin) has to build his local tree _by hand_ 
(or use existing one, if any). 
Now we have such situation: the admin installs the whole stuff on the server.
He has to know perfectly  what more can be done with configuration 
(_rare_ case, believe me). 
If some user wants to try and work with TeX, he _must_ use the global 
configuration (printer, etc.), or he has to learn enought, build his local 
tree _by hand_, copy and modify some files from the global configuration, 
etc. All that rather frustrating, according my mail box.
Conclusion: perhaps it would be wise to set, by default, in texmf.cnf 
someting like:
HOMETEXMF = $HOME/texmf
VARTEXMF = $HOMETEXMF
thus allowing the user starting `texconfing init' on demand.
Believe me, most people do ignore an option of setting VARTEXMF during 
install ;-) More exposed is VARTEXMF.
I would be nice that `texconfig init' will copy also texmf.cnf, what do 
you think, Thomas?

---------

III. AMSFONTS
-------------
In basic installation we have all bsr Type1 files (euler, cyrillic, symbols),
by we haven't tfm files for them.
Perhaps this fonts can be added to lists/free/fonts2/amsfonts
(with all accompanying afm etc. files).
By the way, 
 texmf/fonts/afm/bluesky/euler
 texmf/fonts/afm/bluesky/cyrillic
 texmf/fonts/afm/bluesky/symbol
are missing on TL CD (we have only afm/ams/cmextra/ and afm/ams/cyrillic/)
Well, it is the best time to clean up all that mess.
1. all such fonts are copyrighted (and blessed) by AMS
   please look at, e.g., fonts/afm/bluesky/cm/cmb10.afm  ;-)
                                  =========
2. a part of work was done by Y&Y and a consortium of scientific publishers;
3. now all fonts are freely available for general use, as stated in READ.ME;
4. all fonts reside in fonts/amsfonts/ area on CTAN;
5. I hope that we can do the following:
   texmf/fonts/afm/ams/cmextra/
   texmf/fonts/afm/ams/cyrillic/
   texmf/fonts/afm/ams/euler/
   texmf/fonts/afm/ams/symbol/

   texmf/fonts/pfm/ams/cmextra/
   texmf/fonts/pfm/ams/cyrillic/
   texmf/fonts/pfm/ams/euler/
   texmf/fonts/pfm/ams/symbol/

   texmf/fonts/tfm/ams/cmextra/
   texmf/fonts/tfm/ams/cyrillic/
   texmf/fonts/tfm/ams/euler/
   texmf/fonts/tfm/ams/symbol/

   texmf/fonts/type1/ams/cmextra/
   texmf/fonts/type1/ams/cyrillic/
   texmf/fonts/type1/ams/euler/
   texmf/fonts/type1/ams/symbol/
and we will have finally a consistent tree. 
I know well that this means a little revolution (also in distributions), 
but it is worth of it!
Freezing the actual situation is worse, I mean.
Nevertheless, we have to add to, say, texmf/doc/fonts/amsfonts/type1/
another information about Type1 fonts and their history and copyright  
(CTAN//fonts/amsfonts/ps-type1/READ.ME).

Now, what to do with CM Type1?
a) asking BSR for agreement in such positioning in TDS tree:
   texmf/fonts/afm/public/cm/
   texmf/fonts/pfm/public/cm/
   texmf/fonts/type1/public/cm/
in that case we have consistent tree with fonts/sources/  and fonts/tfm/
(may I ask somebody to pass this problem to Blue Sky Research? I believe
that they will agree. Many thanks in advance.)

b) leaving `as it is':  
   texmf/fonts/afm/bluesky/cm/
   texmf/fonts/pfm/bluesky/cm/ (missing on the last TL)
   texmf/fonts/type1/bluesky/cm/
somehow inconsistent and artificial. 

We can leave all stuff for dvips as is.

---------
I'd also propose: 
 mv texmf/lists/free/latex2/mflogo texmf/lists/free/fonts1/
and change 1 line in it.
I think mflogo are more important then e.g. fonts1/mfmisc.

---------
texmf/dvips/gsftopk/render.ps should be changed to 1.18

---------

That's enouht, I'm a little tired...

Staszek Wawrykiewicz
email: staw@gust.org.pl