required files from unicode.org for jed-extra

[G. Milde]

On 24.07.08, Rafael Laboissiere wrote:
> * G. Milde <milde at users.sourceforge.net> [2008-07-24 12:28]:

ch_table.sl from jed-extra uses ...
> > two description files from the Unicode Character Database:
> >   Blocks.txt
> >   NamesList.txt
...
> > What shall we do:

> > * ship these files with jed-extra

> This would be my preference.  It is way less hassle for us and we can always
> switch to another solution in the future when/if the files go into a
> reasonable location and are provided by a tiny package.

> >   - keep a copy in the upstream archive, or
> >   - download separately with rule get-orig-source

> I would prefer the second option.

OK (if we go this way). 

> That said, if we go down this road, we must check about the DFSG-compliance
> of the Unicode Copyright. I am going to ask this in debian-legal.

This is OK, thanks to Rafael for checking.

But, where should we store the files?

Alternative:

Maybe we can use the method from the command line utility (and Debian
package) "unicode"::

  Package: unicode
  ...
  Suggests: perl-modules | console-data (<< 2:1.0-1) | unicode-data

I.e. jed-extra could 

* Suggest perl-modules | unicode-data 
  (I am not so sure about the suitability of console-data (<< 2:1.0-1))

* Search for the directories

  % /usr/share/unidata/   % (where does this come from?)
  /usr/share/unicode/
  /usr/share/perl/*/unicore/UnicodeData.txt    % with globbing
  ~/.unicode/             % local install, maybe as fallback
  
  which (normally) should contain the needed files.

IMO, this is "the right way". Advantages: 

* No duplicate files on the system.
* Works also for users of the upstream ch_table mode.
* We do not need to care about download, storage, copyright, and updating.

* this would also work (with a modified Suggests:) if eventually
  unicode-data is split into a core and extension package.

Günter