Book logo xindy

A Flexible Indexing System

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

xindy presentation at EuroTeX'98

	Hello out there,

as some of you may have noticed, I gave a talk together with Joachim
about xindy at the EuroTeX'98 conference in St. Malo, March, 30th.

On March, 31st we had a BOF session about xindy where we discussed
several issues how xindy can be further improved and extended. 

The most important results from this session and some other issues are
as follows:

- Improvement of LaTeX support in the future. This mostly concerns a
  LaTeX style file that writes the index entries into the .aux files.
  Additionally, we think about LaTeX macros that write information
  about the document itself (such as encodings and language) into the
  .aux file as well. This enables one to add index style definitions
  into the LaTeX source, instead of a completely separate index style
  file, which seems to be one of the greatest hurdles for the
  acceptance of xindy.
- Support for input filters. Many have asked for such a feature. This
  allows to tag index entries from the raw index with an additional
  atttribute, indicating the index class the index entry belongs to.
  xindy will be extended to accept only those index entries that
  belong to the selected classes.

  My current suggestion is to add a new option :CLASS <string> to the
  INDEXENTRY command to indicate the class of an index entry.
  Additionally, we need another index style command such as
  (INPUT-FILTER (<list-of-class-names>)) that defines the set of class
  names to accept. This specification should be made available to the
  command line options of xindy as well.

- Hans Hagen asked for a way to pass data attachments through the
  indexing process into the tagged index. This would enable xindy to
  process glossaries (with the glossary text as a data attachment) and
  sort the entries with xindy. 

  Several small problems have to be solved to implement such a scheme.
  My suggestion is to add data in the form of a dictionary of
  keyword/value pairs in the raw index interface. The question is
  whether data should be attached to location references or index
  entries or both. Any ideas are welcome.

  This dictionary asks how the mark-up of this data in the backend
  should be implemented. I have some ideas how to realize this, but
  this will need some time.

- Add new hooks in the sorting process to allow sorting rules that can
  be applied only once at the beginning of the rewriting procedure.

- Further simplification of the installation procedure.

- Re-structuring of the documentation.

- many other things to do....

Hope, that gives you an impression what is planned for the future.


Roger Kehr
Computer Science Department         Darmstadt University of Technology