[tex-live] Re: antomega [was: Multilingual LaTeX: Greek, English, and UTF-8]

[Alexej Kryukov]

On Friday 16 September 2005 08:08, Staszek Wawrykiewicz wrote:
>
> or MiKTeX. According to your notes, it is possible to install
> antomega, but only using some same directories already in use, so it
> could break any packaging concept.

First, I am really wondering why the same strict approach is 
not applied to packages like otibet or ocherokee, so that their
files are now installed into tex/lambda/base, although they don't
belong to the official omega distribution. Second (and more important),
I don't think that placing files of a specific package into its
own directory is really always needed to keep "clear distinction"
between packages. On Unix, executable files and libraries are
always installed into the same shared directories, and this doesn't
cause any problems. In texmf tree I would prefer to have similar
shared directories for some kinds of files (see the example with
hyphenation patterns below).

> doc/omega/antomega/
> why not doc/lambda/antomega/ (?)

First, my opinion is that having a directory like doc/lambda is
hardly acceptable by itself. Lambda is an extension for omega, so 
that all lambda specific stuff should go into a common omega tree. 
Thus doc/omega/lambda would be OK; doc/lambda is illogical.

Second, I would accept doc/omega/lambda/antomega as a possible
location for antomega documentation, if antomega was just a
lambda-specific package. However, I have already explained that
some of files provided by antomega can be used with any omega-based
format, and I think (especially after this discussion :) I
really have to provide a better support for plain omega. This
is similar to Babel, which is oriented mainly to latex, but, 
nevertheless, is considered a "generic" package.

So I think doc/omega/antomega is the correct location for
my documentation and should not be changed. Of course the
same reasons are applicable to the 'source' directory.

> omega/hyphen/ -> tex/lambda/antomega/hyphen/
> (or tex/generic/antomega/hyphen in the case that it will work with
> e.g. aleph, or context, as Hans stated?)

Both locations are inacceptable: lambda/antomega/hyphen/ -- because 
this placement would be format specific, while the hyphenation 
patterns are rather compiler specific; generic/antomega/hyphen --
because this directory would be, on the contrary, world-searchable,
i. e. accessible not only for omega and its formats.

In general, I would prefer to have a generic (but compiler-specific)
directory for hyphenation patterns, e. g. tex/omega/generic/hyphen. 
This would be especially reasonable, because, if any hyphenation
patterns are present in the distribution, they should be listed
in the common languages.dat file anyway, so that it is really difficult
to keep distinction between packages here. Also note that in teTeX
there are already some omega-specific hyphenation patterns (cherokee,
inuit...), which surely need a better location than lambda/base.

> omega/lambda/base/

This directory contains one single file, called languages.dat.sample,
and I have already explained that it should be merged with omega's
default languages.dat.

> omega/lambda/config/
> omega/lambda/encodings/
> even if all the above is moved to tex/lambda/, it means that
> everybody can agree that lambda without antomega is unusable, so only
> in such case you can put your files and mix them with lambda. I wrote
> about that...

Encoding descriptions probably can be moved to lambda/antomega.
I'll think over it.

> omega/ocp/antomega/
> OK
>
> omega/ocp/char2uni/
> omega/ocp/uni2char/
> Not consequent with the above! Should be rather
> omega/ocp/antomega/uni2char ?

It *is* consequent. Here I follow the convention introduced by
omega developers themselves, who separated their OCP's (and OTP's as
well) between 3 directories: 'char2uni', 'uni2char' and 'omega'. The
first 2 directories contain files responsible for standard
charset-to-charset conversion. The last directory is package-specific:
the OCP's stored here are used to perform ligature substitutions or
similar things. These files may also produce TeX/Omega command
sequences in their output, so that they can be used only by package
where these command sequences are defined.

Note that this later directory is named 'omega' not because of
its relation with the omega compiler (this would be trivial), but
because it was used by the omega.sty package until it was broken.
That's why I have also distributed my OCP's between 'char2uni',
'uni2char' and 'antomega'. If you wish, I can move everything in
omega/ocp/char2uni/ to omega/ocp/char2uni/antomega (and the same
thing for omega/ocp/uni2char), but placing these files into
omega/ocp/antomega would be incorrect.

> omega/unidata/
> The same as above (not consequent).

Like the hyphenation patterns, this directory is compiler-specific, 
not format-specific. An acceptable location would be something like 
tex/omega/generic/antomega/unidata, but exact placement of
compiler-specific files is still subject to discussion.

-- 
Regards,
Alexej Kryukov <akrioukov at newmail dot ru>

Moscow State University
Historical Faculty