Re: texconfig action and site-wide changes

2005-02-09 Thread Hartmut Henkel
On Tue, 8 Feb 2005, George N. White III wrote:

> On Tue, 8 Feb 2005, Thomas Esser wrote:
>
> > > Did you notice that the table of contents of the INSTALL file is
> > > still:
> >
> > Yes, I noticed it when I looked it up today. Unfortunately, I did
> > not notice it before the release. :-(
>
> Nor any of the testers!  Microsoft has the advantage that they can
> hire someone who has never used a product

they have lots of testers actually _paying_ for it (called "users").

> and ask them to try installing it during testing.

SCNR, Regards, Hartmut


Re: texconfig action and site-wide changes

2005-02-09 Thread David R. Morrison
On Feb 8, 2005, at 2:08 PM, Thomas Esser wrote:
Did you notice that the table of contents of the INSTALL file is 
still:
Yes, I noticed it when I looked it up today. Unfortunately, I did not
notice it before the release. :-(
Thomas
While we're all belatedly proofreading the INSTALL file: I've just 
noticed that, in section 3, you probably want an extra line in the 
standard directory layout:

  $HOME/texmf % user tree for added fonts and macros
(since this seems to be present by default in texmf.cnf).
  -- Dave


Re: texconfig action and site-wide changes

2005-02-08 Thread George N. White III
On Tue, 8 Feb 2005, Thomas Esser wrote:
Did you notice that the table of contents of the INSTALL file is still:
Yes, I noticed it when I looked it up today. Unfortunately, I did not
notice it before the release. :-(
Nor any of the testers!  Microsoft has the advantage that they can hire 
someone who has never used a product and ask them to try installing it 
during testing.  TeTeX testers tend to be experienced so most probably 
didn't bother reading INSTALL.  It would be nice to get a few people who 
are new to TeX involved in testing.

--
George N. White III  <[EMAIL PROTECTED]>


Re: texconfig action and site-wide changes

2005-02-08 Thread George N. White III
On Tue, 8 Feb 2005, Thomas Esser wrote:
What would help is some tools to check for duplicates across all the
texmf trees so at least you can more easily determine when things in
$TEXMFHOME or $TEXMFLOCAL have become older than versions in updated
system trees.
Such as the one mentioned in Appendix G of the INSTALL document?
   =
G) scanning texmf trees for duplicates
   
Yes, although it would help to put a separator between the 'ls -l' output
and the md5sum, e.g.:
  $info =~ s/ ([^ ]+$)/$md $1/;
--
George N. White III  <[EMAIL PROTECTED]>


Re: texconfig action and site-wide changes

2005-02-08 Thread Thomas Esser
> Did you notice that the table of contents of the INSTALL file is still:

Yes, I noticed it when I looked it up today. Unfortunately, I did not
notice it before the release. :-(

Thomas


Re: texconfig action and site-wide changes

2005-02-08 Thread Frank Küster
Thomas Esser <[EMAIL PROTECTED]> schrieb:

>> What would help is some tools to check for duplicates across all the
>> texmf trees so at least you can more easily determine when things in
>> $TEXMFHOME or $TEXMFLOCAL have become older than versions in updated
>> system trees.  
>
> Such as the one mentioned in Appendix G of the INSTALL document?
>
> 
> =
>  G) scanning texmf trees for duplicates
> 
> =
> ...

Did you notice that the table of contents of the INSTALL file is still:

,
|   [...]
|   7) final configuration steps
|   A) appendix: notes on some platforms
|   B) appendix: note on moving the binaries around
|   C) appendix: note on updating a single applications
|   D) appendix: recreating format files
| 
| 
`

appendices E to G are not mentioned.

Regards, Frank
-- 
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer



Re: texconfig action and site-wide changes

2005-02-08 Thread Thomas Esser
> What would help is some tools to check for duplicates across all the
> texmf trees so at least you can more easily determine when things in
> $TEXMFHOME or $TEXMFLOCAL have become older than versions in updated
> system trees.  

Such as the one mentioned in Appendix G of the INSTALL document?


=
 G) scanning texmf trees for duplicates

=
...

:-)

Thomas


Re: texconfig action and site-wide changes

2005-02-08 Thread George White
On Tue, 8 Feb 2005, Thomas Esser wrote:

> > I have tried to understand how texconfig works if a user invokes it. As
> > far as I can see, for every configuration file they touch, a copy is
> > generated in TEXMFCONFIG (which is $HOME/.texmf-config by default).
> 
> Right.
> 
> > This seems to have the (probably unwanted effect) that the user is thus
> > cut-off from site-wide changes.
> 
> The same is true if some user puts a custom copy of koma-script into
> his $TEXMFHOME.
> 
> > If a user changes a configuration file $cfile for the first time, the
> > changed file after "check_out" is not only copied or cat'ed to
> > $TEXMFCONFIG/$relDir/$cfile, but additionally a diff or the change regex
> 
> Well, this sounds like a complicated solution (different config files
> would need different kind of updates) with a questionable effect.

What would help is some tools to check for duplicates across all the
texmf trees so at least you can more easily determine when things in
$TEXMFHOME or $TEXMFLOCAL have become older than versions in updated
system trees.  

--
George White <[EMAIL PROTECTED]> <[EMAIL PROTECTED]>
189 Parklea Dr., Head of St. Margarets Bay, Nova Scotia  B3Z 2G6


Re: texconfig action and site-wide changes

2005-02-08 Thread Thomas Esser
Hi Frank,

> It seems to me that "texconfig-sys init" should call fmtutil-sys and
> updmap-sys, too, not fmtutil and updmap plain - I didn't check for
> texlinks.]

That is wrong, just try
  texconfig-sys formats
change some bits in the config file (->goes via "fmtutil --edit" to
TEXMFSYSCONFIG) and the changed/added formats to via "fmtutil --byfmt"
to TEXMFSYSVAR).

The environment manipulation done in texconfig-sys has of course not
only some effect on the texconfig script that it calls, but also to all
other subprocesses.

> I have tried to understand how texconfig works if a user invokes it. As
> far as I can see, for every configuration file they touch, a copy is
> generated in TEXMFCONFIG (which is $HOME/.texmf-config by default).

Right.

> This seems to have the (probably unwanted effect) that the user is thus
> cut-off from site-wide changes.

The same is true if some user puts a custom copy of koma-script into
his $TEXMFHOME.

> If a user changes a configuration file $cfile for the first time, the
> changed file after "check_out" is not only copied or cat'ed to
> $TEXMFCONFIG/$relDir/$cfile, but additionally a diff or the change regex

Well, this sounds like a complicated solution (different config files
would need different kind of updates) with a questionable effect.

Thomas


texconfig action and site-wide changes

2005-02-07 Thread Frank Küster
Hi all,

[Something completely different, which I found while looking at this:

It seems to me that "texconfig-sys init" should call fmtutil-sys and
updmap-sys, too, not fmtutil and updmap plain - I didn't check for
texlinks.]


I have tried to understand how texconfig works if a user invokes it. As
far as I can see, for every configuration file they touch, a copy is
generated in TEXMFCONFIG (which is $HOME/.texmf-config by default).

This seems to have the (probably unwanted effect) that the user is thus
cut-off from site-wide changes. For example, if he changes updmap.cfg
because he has an additional font in TEXMFHOME, and afterwards the site
administrator adds a second font to TEXMFLOCAL and edits the site-wide
updmap.cfg accordingly, this change will not propagate to the user.

It seems to me that it would be a good idea to try to implement a scheme
for texconfig action in which user changes could override particular
site-wide settings, but would still use the settings in site-wide files
if possible.  I am not sure that this is feasible, but at least I wanted
to state my wish.


Maybe it could be done along the following lines:

If a user changes a configuration file $cfile for the first time, the
changed file after "check_out" is not only copied or cat'ed to
$TEXMFCONFIG/$relDir/$cfile, but additionally a diff or the change regex
is saved in $TEXMFCONFIG/$relDir/$cfile.diff. If the user invokes
texconfig again to change the same file, tcmgr will not check out the
changed file (i.e. the file kpsewhich would find), but rather first take
the file from SYSTEXMF, apply the diff or ReplaceRegex and use the
result as the checked out file (it could also check for differences in
SYSTEXMF and report those changes to the user). A command "texconfig
update" would then be handy to merge site-wide changes into all existing
user-specific configuration files.

For some configuration files, it might be even easier. texmf.cnf would
be an example - just that it cannot be accessed via texconfig AFAIK. But
for this file, it would be sufficient to have *only* the lines the user
wants changed in his file. libkpathsea will read all available texmf.cnf
files, first the user's and then the site-wide, and the first
definitions will take precedence. Maybe a similar behavior could be
implemented for other configuration files, too. I only fear that it
cannot be applied in all cases, because I don't see how one could remove
settings completely in this way.

Regards, Frank

-- 
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer