Re: 2.99.11: yet another release candidate

2005-02-04 Thread Thomas Esser
> For Cygwin, I have been using using the following commands during
> postinstall
> 
> unset MFINPUTS TEXINPUTS TEXMF TEXMFCNF TFMFONTS VARTEXFONTS
> 
> texconfig rehash
> texconfig confall
> texconfig rehash
> texconfig init
> 
> # default dvips output to file
> texconfig dvips printcmd -
> 
> to create the initial formats.  Can these just be replaced by
> texconfig-sys,


Yes, you should use texconfig-sys. It should be sufficient to run
  mktexlsr
  texconfig-sys init
  texconfig-sys dvips printcmd -

> and will that then update as before to the main tree?

No, this will not update $TEXMFMAIN. The map files / formats will be
written to TEXMFSYSVAR.

> (I'll test with .12 if I get the chance.)

Please, check if the settings for TEXMFVAR / TEXMFCONFIG make sense
on Cygwin.

Thomas


Re: 2.99.11: yet another release candidate

2005-02-04 Thread Erik Frisk
Thanks Thomas for all your feedback to my questions. Now everything 
(20050203) works fine both on my Linux FC3 box at home and in my Solaris 
environment at work (although I have not yet fully understood the 
principles behind all texmf-trees).

Thanks for a great distribution!
/Erik
--
-
Erik Friskemail: [EMAIL PROTECTED]
Dept. of Electrical Engineering tel:+46 13 285714
Linköping Universityfax:+46 13 282035
S-581 83 Linköping
SWEDEN   http://www.fs.isy.liu.se/~frisk/
-


Re: 2.99.11: yet another release candidate

2005-02-04 Thread Jan Nieuwenhuizen
Thomas Esser writes:

> The exciting new stuff this time are the new defaults for TEXMFCONFIG
> / TEXMFVAR which are in the users's home directories. So, the tools
> texconfig, fmtutil and updmap can be used by all users in a multi-user
> setup. For changing the global defaults, -sys variants of the above
> commands have been created (texconfig-sys, fmtutil-sys, updmap-sys).

..

> So, formats, map files etc. no longer end up in TEXMFMAIN.

For Cygwin, I have been using using the following commands during
postinstall

unset MFINPUTS TEXINPUTS TEXMF TEXMFCNF TFMFONTS VARTEXFONTS

texconfig rehash
texconfig confall
texconfig rehash
texconfig init

# default dvips output to file
texconfig dvips printcmd -

to create the initial formats.  Can these just be replaced by
texconfig-sys, and will that then update as before to the main tree?

(I'll test with .12 if I get the chance.)

Jan.

-- 
Jan Nieuwenhuizen <[EMAIL PROTECTED]> | GNU LilyPond - The music typesetter
http://www.xs4all.nl/~jantien   | http://www.lilypond.org


Re: 2.99.11: yet another release candidate

2005-02-01 Thread Thomas Esser
Hi Erik,

thanks for your testing and feedback!

> I am not sure where generated font-files are best placed, but my 
> understanding is that $TETEXDIR/share/texmf should hold any 
> automatically generated stuff like format files and font files.

The default is to have the font files in /var/tmp/texfonts. If you want
to use the texmfmain tree for that, you should:

  texconfig-sys font vardir $TETEXDIR/share/texmf/fonts
  mkdir $TETEXDIR/share/texmf/fonts
  chmod 1777 $TETEXDIR/share/texmf/fonts

That way, the fonts directory (and any automatically created subdirectory)
will be writable by everybody.

I might add something like this to the documantation or add support for
it to texconfig.

> Since I have failed to see the reason for a $TETEXDIR/share/texmf-var 
> tree when I already have the $TETEXDIR/share/texmf tree, I have omitted 
> the texmf-var tree.

Still, you *have* to set TEXMFSYSVAR and TEXMFSYSCONFIG in your texmf.cnf
file. You can assign the same value taht TEXMFMAIN has, but you have to
set it.

> I haev also always wondered how to manage write permission for the font 
> directories.

It used to be simple. Now, we have $TEXMFMAIN/web2c/mktexdir which
uses kpsestat. To some extend, permissions of the parent directory are
inherited (and a sticky bit might be added).

> feels highly unsatisfactory. What is the "right" way to do it?

Set the right permissions to the parent directory.

> However, if I then do 'fmtutil-sys --all', format files still get 
> generated in ~/.texmf-var, not in $TETEXDIR/share/texmf as I wanted (or 
> $TETEXDIR/share/texmf-var as I had expected)

I see that /bin/sh on solaris does not put a=b into the environment if
exec is used in the following way:
  /bin/sh -c 'a=b exec env'

I have not expected this and I guess that on most systems with a standard
bourne shell this makes all the -sys variants useless. Sorry! (I am
grateful that this was detected that early!)

Try to use
  TEXMFVAR="$v"; TEXMFCONFIG="$c"; export TEXMFVAR TEXMFCONFIG; exec fmtutil 
${1+"$@"}
instead of
  TEXMFVAR="$v" TEXMFCONFIG="$c" exec fmtutil ${1+"$@"}
(same change for texconfig-sys and updmap.sys).

> Can someone enlighten me how to set up the environment.

It is not your fault!

> Also on a final note, I have the cm-super fonts installed in the local 
> texmf-tree and would like to enable them. Previously I have edited 
> updmap.cfg by hand but now I tried to use the updmap-sys script like this:

  updmap-sys --edit
does what you want.

...
> updmap-sys --enable Map=cm-super-x2.map
> 
> This does not seem to work as expected.

It does not? How does it fail? What is the error/problem?

Thomas


Re: 2.99.11: yet another release candidate

2005-02-01 Thread Erik Frisk
Hi all.
I have not yet grasped all new changes and am currently a little 
confused about all texmf-trees and environment variables. So I explain 
my confusion and hopfully someone can set me straight(er).

Goal

I have the following structure on my filesystem for the tetex installation
.../tex/teTeX/20050131 # $TETEXDIR
.../tex/local/texmf# local texmf-tree
In my local texmf-tree the file .../tex/local/texmf/web2c/texmf.cnf
consists of the lines:
TEXMFLOCAL=.../tex/local/texmf
TEXMF = 
{$HOMETEXMF,!!$TEXMFLOCAL,!!/home/fs/frisk/usr/tex/local/texmf.tex4ht,!!$VARTEXMF,!!$TEXMFMAIN,!!$TEXMFDIST}

The goal is a tex-installation for the department with a local texmf 
tree that includes some style-files and possibly also some font-files.

I am not sure where generated font-files are best placed, but my 
understanding is that $TETEXDIR/share/texmf should hold any 
automatically generated stuff like format files and font files.
Since I have failed to see the reason for a $TETEXDIR/share/texmf-var 
tree when I already have the $TETEXDIR/share/texmf tree, I have omitted 
the texmf-var tree.

I haev also always wondered how to manage write permission for the font 
directories. Since, if a user uses a font for the first time, a new 
directory is created with permissions only for that user. The next user 
who uses that font can now not generate files in that directory. I have 
up til now "solved" this by doing recurrent 'chmod -R' calls. But this 
feels highly unsatisfactory. What is the "right" way to do it?

Installation procedure
==
1) configure --with-xdvi-x-toolkit=xaw --prefix=$TETEXDIR
   make world
2) texconfig-sys font vardir $TETEXDIR/share/texmf
Post-installation procedure
===
Set environment variable $TEXMFCNF to .../tex/local/texmf/web2c/:
However, if I then do 'fmtutil-sys --all', format files still get 
generated in ~/.texmf-var, not in $TETEXDIR/share/texmf as I wanted (or 
$TETEXDIR/share/texmf-var as I had expected)

Now, when reading the release notes I see that there are also 
texmf-config trees. I also feel uncertain how the user texmf-trees 
(~/.texmf-var, ~/.texmf-config etc.) interact with the system trees.

Can someone enlighten me how to set up the environment.
Also on a final note, I have the cm-super fonts installed in the local 
texmf-tree and would like to enable them. Previously I have edited 
updmap.cfg by hand but now I tried to use the updmap-sys script like this:

updmap-sys --enable Map=cm-super-t1.map
updmap-sys --enable Map=cm-super-ts1.map
updmap-sys --enable Map=cm-super-t2a.map
updmap-sys --enable Map=cm-super-t2b.map
updmap-sys --enable Map=cm-super-t2c.map
updmap-sys --enable Map=cm-super-x2.map
This does not seem to work as expected. Also, is there a way to specify 
all new map-files in a single call to updmap-sys?

Thanks for any help,
Erik
-
Erik Friskemail: [EMAIL PROTECTED]
Dept. of Electrical Engineering tel:+46 13 285714
Linköping Universityfax:+46 13 282035
S-581 83 Linköping
SWEDEN   http://www.fs.isy.liu.se/~frisk/
-


Re: 2.99.11: yet another release candidate

2005-02-01 Thread David Kastrup
Willi Burmeister <[EMAIL PROTECTED]> writes:

> Hi,
>
>> > I like this feature, but what about calling -sys variants if the
>> > main tree is writeable by the current user?
>> 
>> Why should one do that?  If a particular user is maintaining the TeX
>> tree, it is important that he has the ability to do maintainance both
>> for himself as well as the system.
>
> We have a special user 'tex', created only to maintain our TeX tree.
> This is not a real user and so there is no need to maintain 'his' user
> profile.
>
> We have special user for many global software installations. This makes
> it possible to change maintainer without dependencies.

So use texconfig-sys for this, or set up the environment variables
accordingly.  You'll have to do similar things for every other
software installation, too.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum


Re: 2.99.11: yet another release candidate

2005-02-01 Thread Willi Burmeister
Hi,

> > I like this feature, but what about calling -sys variants if the
> > main tree is writeable by the current user?
> 
> Why should one do that?  If a particular user is maintaining the TeX
> tree, it is important that he has the ability to do maintainance both
> for himself as well as the system.

We have a special user 'tex', created only to maintain our TeX tree.
This is not a real user and so there is no need to maintain 'his' user
profile.

We have special user for many global software installations. This makes
it possible to change maintainer without dependencies.

Willi



Re: 2.99.11: yet another release candidate

2005-01-31 Thread David Kastrup
Willi Burmeister <[EMAIL PROTECTED]> writes:

>> The exciting new stuff this time are the new defaults for
>> TEXMFCONFIG / TEXMFVAR which are in the users's home
>> directories. So, the tools texconfig, fmtutil and updmap can be
>> used by all users in a multi-user setup. For changing the global
>> defaults, -sys variants of the above commands have been created
>> (texconfig-sys, fmtutil-sys, updmap-sys).
>
> I like this feature, but what about calling -sys variants if the
> main tree is writeable by the current user?

Why should one do that?  If a particular user is maintaining the TeX
tree, it is important that he has the ability to do maintainance both
for himself as well as the system.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum


Re: 2.99.11: yet another release candidate

2005-01-31 Thread Thomas Esser
> Nevertheless, a clear warning when the user-specific texconfig is run
> by root seems appropriate: if people set up root-specific defaults for

I hate to make lots of special cases (writable directories, user is root etc.).

To consider:
  - texconfig-sys calls texconfig, so the invocation of texconfig would
need to know that root has called texconfig-sys in the first place
and not texconfig directly
  - consider the chain texconfig-sys -> texconfig -> fmtutil; sure,
you don't want that fmtutil issues a warning.
  - root might have read the documentation and set TEXMFVAR / TEXMFCONFIG
to something that is useful
  - making texconfig to prompt for something might break something else
  - a warning without a prompt almost useless (who reads all the messages
that pass along!?)
  - texconfig is not the only tool to consider

So, I don't think that I should do that.

Thomas


Re: 2.99.11: yet another release candidate

2005-01-31 Thread Thomas Esser
> user types "texconfig init", producing these many formats. Is there a
> trick that the local texmf.cnf is compared to the global one, and just
> the formats for differing lines are generated?

Well, as administrator for a large multi-user site, you can support your
users by writing something like this (copied from texconfig):

comm -13 $tmpdir/pre $tmpdir/post > $tmpdir/addOrChange
for i in `awk '{print $1}' $tmpdir/addOrChange`; do
  fmtutil --byfmt "$i" || rc=1
done

$tmpdir/pre would be the "global" file (sorted, without comments),
$tmpdir/post the user's file (sorted, without comments). comm finds the
changed/added lines and awk prints the names of the formats that are
affected. fmtutil --byfmt being called for each of those format names
rebuilds them.

I don't have something ready-to-use right now, but this should give you
an idea about how it can be done.

Thomas


Re: 2.99.11: yet another release candidate

2005-01-31 Thread Willi Burmeister
Hi Thomas,

> The exciting new stuff this time are the new defaults for TEXMFCONFIG
> / TEXMFVAR which are in the users's home directories. So, the tools
> texconfig, fmtutil and updmap can be used by all users in a multi-user
> setup. For changing the global defaults, -sys variants of the above
> commands have been created (texconfig-sys, fmtutil-sys, updmap-sys).

I like this feature, but what about calling -sys variants if the main 
tree is writeable by the current user?

> As usual: please, send me feedback, even if it all just works :-)

2.99.11 compiled without a problem on my Solaris 9 Ultra 60 (Sun C 5.5).
Same result on our V20z with Solaris Express (Sun C 5.5).

Some quick tests with texconfig and several tex files didn't show any
problem.

> If you have time to read the documentation (yes, I have updated parts of 
> INSTALL,

Please add 

  - Sparc/Solaris 9, Sun C 5.5, GNU flex 2.5.4, GNU bison 1.875, GNU make 3.80
  - x86/Solaris 10, Sun C 5.5, GNU flex 2.5.4, GNU bison 1.875, GNU make 3.80

in appendix A :-)


Willi



Re: 2.99.11: yet another release candidate

2005-01-31 Thread Hartmut Henkel
It's not fully clear to me how user-side fmtutil.cnf is handled.
Traditionally i (user) add multilanguage support to tex and pdftex (i
don't like to type bplain :-). After editing fmtutil.cnf, these two
formats are generated in local .texmf-var. Ok. But when a global format
changes, the local user tex/pdftex is stymied. User has to selectively
re-generate the "stymied" formats e. g. by "texconfig init pdftex". But
when user types "texconfig init", the whole bunch of format files is
re-generated in his home directory, unnecessarily. Actually i want only
the ones that differ in the texmf.cnf from the global texmf.cnf (if the
machine would be a big multi-user server, after some time many users
would end up with their bunch of formats in their home directories, as
"texconfig init" is so easy to type).

When i remove all files from local .texmf-var/web2c and then go through
texconfig's format menue, the state is analyzed but no local format is
generated. Actually the global formats are used, maybe contradicting
what's in the local texmf.cnf. Only direct way to be safe seems that a
user types "texconfig init", producing these many formats. Is there a
trick that the local texmf.cnf is compared to the global one, and just
the formats for differing lines are generated?

Well, i'm already happy that a simple "texconfig init pdftex" works
fine.

And, maybe a way out would be that the local user only gets an
fmtutil.cnf where all lines are simply commented out by default. Then
the user can remove comments only for the formats he really would like
to have locally. And then he indeed could simply type "texconfig init".

Regards, Hartmut


Re: 2.99.11: yet another release candidate

2005-01-31 Thread David Kastrup
Hartmut Henkel <[EMAIL PROTECTED]> writes:

> On Mon, 31 Jan 2005, Thomas Esser wrote:
>
>> The exciting new stuff this time are the new defaults for TEXMFCONFIG
>> / TEXMFVAR which are in the users's home directories.
>
> this time i was rather shocked: things suddenly going into hidden
> /root/.texmf... Then i re-read the announcement and tried texconfig in
> my home directory, and it just works fine :-) Very nice concept!
>
>> but everything is clearly separated. I like it.
>
> yes it is.

Nevertheless, a clear warning when the user-specific texconfig is run
by root seems appropriate: if people set up root-specific defaults for
TeXing, chances are that they are doing something imprudent,
intentionally or not.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum


Re: 2.99.11: yet another release candidate

2005-01-31 Thread Hartmut Henkel
On Mon, 31 Jan 2005, Thomas Esser wrote:

> The exciting new stuff this time are the new defaults for TEXMFCONFIG
> / TEXMFVAR which are in the users's home directories.

this time i was rather shocked: things suddenly going into hidden
/root/.texmf... Then i re-read the announcement and tried texconfig in
my home directory, and it just works fine :-) Very nice concept!

> but everything is clearly separated. I like it.

yes it is. No problems at all (debian woody i686). I like it, too.

Regards, Hartmut


Re: 2.99.11: yet another release candidate

2005-01-31 Thread VnPenguin
On Mon, 31 Jan 2005 00:28:26 +0100, Thomas Esser <[EMAIL PROTECTED]> wrote:
> This 2.99.11 is the fifth (and hopefully last) release candidate of
> teTeX-3.0. Again, there have been several changes (see below) which I'd
> like to see tested.
> 
...

Just build 2.99.11. Compiling is ok on my FC3 Linux box.
But when I try to compile my test (vntex, utf-8 input encoding)  with
ConTeXt I have error:

/work/apps/tetex/share/texmf/web2c/mktexnam: Could not map typeface
abbreviation 5- for t5-utmr8a.
/work/apps/tetex/share/texmf/web2c/mktexnam: Need to update
/work/apps/tetex/share/texmf-dist/fonts/map/fontname/special.map?
mktextfm: Running mf-nowin -progname=mf \mode:=ljfour; mag:=1;
nonstopmode; input t5-utmr8a
This is METAFONT, Version 2.71828 (Web2C 7.5.4)

kpathsea: Running mktexmf t5-utmr8a
! I can't find file `t5-utmr8a'.
<*> ...jfour; mag:=1; nonstopmode; input t5-utmr8a
  
Please type another input file name
! Emergency stop.
<*> ...jfour; mag:=1; nonstopmode; input t5-utmr8a
  
Transcript written on mfput.log.
grep: t5-utmr8a.log: No such file or directory
mktextfm: `mf-nowin -progname=mf \mode:=ljfour; mag:=1; nonstopmode;
input t5-utmr8a' failed to make t5-utmr8a.tfm.
kpathsea: Appending font creation commands to missfont.log.

! Font \*postscript12ptrmtf*:=t5-utmr8a at 12.0pt not loadable: Metric (TFM) fi
le not found.
 
   \relax 
\xxdododefinefont ...tspec {#4}\newfontidentifier 
  \let \localrelativefontsiz...
\fontstrategy ...me \fontclass #2#3#4#5\endcsname 
  \tryingfontfalse \fi 
 ...yle \fontalternative \fontsize 
  \fi \iftryingfont \fontstr...

\synchronizefont ...strategy \the \fontstrategies 
  \relax \fi \iftryingfont \...
 \getvalue [EMAIL PROTECTED]@ \fontstyle }
   \edef \fontstyle {\fontstyle }\if...
...
l.140 \setupbodyfont[postscript,12pt]

-
With 2.99.10 I could compile without problem with the same src.
When I do research in my TEXMF:

# find . -name "*utmr8*"
./texmf-dist/fonts/afm/urw/times/utmr8a.afm
./texmf-dist/fonts/tfm/public/urwvn/utmr8v.tfm
./texmf-dist/fonts/type1/urw/times/utmr8a.pfb
./texmf-dist/fonts/type1/public/urwvn/utmr8v.pfb

Any ideas for help ?

Thank you,

Quy NGUYEN-DAI


Re: 2.99.11: yet another release candidate

2005-01-31 Thread Artemio Gonzalez Lopez
On Jan 31, 2005, at 12:28 AM, Thomas Esser wrote:
As usual: please, send me feedback, even if it all just works :-) If 
you
have time to read the documentation (yes, I have updated parts of 
INSTALL,
TETETXDOC, teTeX-FAQ tec.), please let me know if the descriptions are
good enough or not.
Everything works so far on Mac OS X 10.3.7.
Thanks a lot,
Artemio Gonzalez-Lopez
[EMAIL PROTECTED]


Re: 2.99.11: yet another release candidate

2005-01-30 Thread Reinhard Kotucha
> "Thomas" == Thomas Esser <[EMAIL PROTECTED]> writes:

  > I have not finally decided about texi2html (should be easy enough
  > to switch to a different version). And I don't know when
  > pdftex-1.20c will be released. Well, the pdftex-beta we have is
  > basically a fixed 1.20b and the fixes seem to work well, so that
  > should be good enough for the 3.0 release.

If there is no possibility to include the latest pdftex, it would be
nice if you could at least include the /UniqueID patch.

Regards,
  Reinhard

-- 

Reinhard Kotucha  Phone: +49-511-4592165
Marschnerstr. 25
D-30167 Hannover  mailto:[EMAIL PROTECTED]

Microsoft isn't the answer. Microsoft is the question, and the answer is NO.