Fixing Broken Stuff
-------------------
* Fix bugs in the bug system: 
    https://sourceforge.net/tracker/?group_id=11543&atid=111543
* Change temporary and log files in extract.c to follow @localstatedir@
* If somebody does a scrollkeeper-install of a document which is already installed,
  it installs a second copy.  It should really uninstall it and then do the install.
* If the identifier url is incorrect, an error is sent to STDOUT but not to the
  log.  Also, the document is added to scrollkeeper_docs even though the document
  is not properly registered.
* Update tarball install to have new features of Jan 16, 2002 RPM updates
* ScrollKeeper should check that a given DTD is installed before trying to
  apply a stylesheet to it, or it will fail.  When this is done, we can
  consider taking the explicit requirement out of configure.in.

Adding/Enriching Features
-------------------------
* Add appropriate flags to various commands to allow for easy use by non-root
  users.  This means each command must accept a flag for the database path,
  and possibly for a log file and other files.
* Add --help flags to each script/binary and update man pages
* Try to capture any errors which libxml2 dumps and send them to the log file
  instead of STDOUT
* Write DTDs for the ScrollKeeper XML data files
* Write a stylesheet to convert the category list to a more readable format,
  such as ASCII or HTML.
* Write a script which finds all sectioning (and maybe some others?) tags
  and inserts id's if they are missing. eg: "scrollkeeper-addids <infile>
  <outfile>".  It just runs xsltproc with a stylesheet that adds a generic
  and reproducable idref. Sections are given ids "genericsectionid######",
  indexing tags are id'd "genericindexid#####". We need this as a pre-processor
  b/c DocBook doesn't require id's for most things, but ScrollKeeper/help browsers
  need them in order to link from the table of contents to particular sections or
  for the index to link to the terms they list.
* See if there is a more graceful way to handle docs where the OMF file
  mis-declares their mime type. Right now, it causes a nasty spew to the screen.
* Add a -q (--quiet) flag for use in the %post scripts.  When -q is used, all
  text output except the most serious errors should not go to STDOUT.  This way
  people don't see any output from ScrollKeeper when installing a package with
  a bad document or OMF file.  Then, we can change various scrollkeeper-* to 
  output some useful info by default instead of being very quiet all the time.
* Generally, we still need to improve our error logging and general info
  logging a lot.  For example, it currently sends a log entry when it tries
  to install or uninstall something.  However, it doesn't distinguish between
  successful and failed attempts.
* Put back the SGML index extraction and TOC extraction, doing proper checking
  for failure of sgml2xml.  It should not try to do the extractions if sgml2xml
  returns errors.

Documentation
-------------
* Write an OMF-writing tutorial, and add it to the web page as well as to the
  doc/ directory so it is installed.  It should be in DocBook/XML.
* Put the category list on the web page.
* Update/finish 0.4 specifications.
* Write a document about anything with DocBook/XML or DocBook/SGML that
  ScrollKeeper is picky about.  For example, ScrollKeeper currently requires that
  a section have an id to be included in the TOC (since it needs an anchor to
  link to).
* Write a FAQ. (1) Q: How does one find out if a specific document is installed, and where? A: Use the contents list and look for <seriesid> to match that of the document of interest.

Testing
-------
* Go forth and test, especially in with various documents (SGML and XML) 
  and with various locales.
* Make sure that docs with invalid categories are placed into the most 
  appropriate "Other" section.
* Make sure that documents with missing document id's all get treated as 
  unique documents.
* Test locale fallback for documents

