0.13
   added "stream context" support to rippy_readfile and higher-level
      routines; this allows use of HTTP proxies, and user agent setting
   improved parsing of unclosed tags when scraping old non-XML HTML
   fixed handling of failed self-modifying format commands; this has the
      effect of fixing "empty URL" bug in default item format
   added sanity check for top-level tag in feed file; this fixes crash when
      delinquent feeds return HTML instead
   minor syntax cleanup

0.12
   added @ to reduce the error messages during installation
   tweaked the conditions for cache updating
   fixed a longstanding bug in XML parsing that affected attribute values
   added translations for a few &#numeric entities
   more cautious clearing of the "recent" flag, to prevent seeing an entire
      feed as new when it comes back from having been broken.
   added rippy_export_opml()
   added Skippy scraping extension - alpha code, not fully documented

0.11
   added "nofollow" support, and commentary for it

0.10
   added per-feed "key" option to allow for title-less entries, etc.
   removed user-controlled filter dependence on register_globals (which
      probably shouldn't be turned on if allow_url_fopen, which the caching
      code needs, is also turned on - SECURITY ISSUE)
   added per-feed retention and display time limits and counts
   added global config parameter for retention and display limits
   fixed keep-everything-forever bug (tsk, that one should never have
      existed) and several secondary bugs created or revealed by this fix
   added "extended" formatting commands:  FILTERIN and FILTEROUT (and the
      "kill" formatting variable) for regexp-based item filtering;
      FIRSTCHOICE (support of the existing formatting system in the new
      framework); HTMLESCAPE (used for RSS output); and REPLACE (to edit
      items, for instance to remove <![CDATA[..]]> constructs).
   added experimental Atom support
   added UTF-8 translation kludge
   added RSS output demo (achwood-blogs.php)
   changed caching to load feed files in random order, so aggregation will
     still work when there are slow feeds on the list, as long as there are
     enough page loads in total.

0.9
   changed filtering code to display items still in the RSS, even if they
      are old, if we're low on items to display (makes aggregator less
      likely to run dry)
   changed filtering code to only display items from feeds that are in the
      feeds array (even if others are in the aggregation file) to permit
      the next feature to work and solve some startup issues
   added optional "display only one or a few feeds" feature, managed by form
      variable, to aggregate.php
   added support for image caching
   the usual addition and removal of default feeds
   moved reading of files (both local and remote) into rippy_readfile() and
      built in my own timeout; this fixes problem noted at baremetal.com
      where Rippy sleeps forever waiting on a delinquent feed, eventually
      filling process table, and neither PHP nor Apache timeouts kill it
   file locking procedures revised and expanded
   change to rippy_updatecache: don't overwrite on failed HTTP transaction

0.8
   fixed inconsistent version number in page title (minor typo)
   added "rippy_" prefix to function names to solve name clash with
      aggregate() built-in function of PHP 4.2 and future problems of this
      type - WARNING - this means you can't upgrade by just dropping in the
      new rippy,php, but must actually edit your calling code
   added translation for &apos; character entity
   fixed Ozy&Millie scraper script
   added clarifications to the README

0.7
   new "formatting" system, allows comprehensive redefinition of the
      generated HTML without touching rippy.php (should ease future upgrades
      for the majority of users who customize the code)
   per-feed information placed in an array of arrays to allow more detailed
      configuration control
   cache updating moved into aggregate() (although you can still do it the
      old way if you really want to); now a new feed only needs to be added
      in one place in the code, although you still have to think about
      creating the cache file.
   more new feeds in the default list, some delinquent ones commented out
   reduced the emphasis on gDefaultIconURL because people hated it

0.6
   made it process all new items in a given update at the same virtual time,
      to fix a bug that would mess up the order of items if there were many
      new in the same file.
   repaired the Ozy&Millie scraper script to accomodate changes on that site
   two changes suggested by Ewen McNeill:
      - changed HTML to be more XHTML-compliant (probably still not perfect)
      - only display <hr /> when there is a description (see comment in code)

0.5
   reorganized source into multiple files for better maintenance/upgrading
   included copies of my comic-scraping scripts
   provided defaults for feeds which lack per-feed icon, title, link

0.4
   added some new default feeds
   learned how to do pass-by-reference in PHP, and used that to move the
      basic parsing operation into a function
   improved parser to deal with RSS 1.0 - it's still a kludge, but now
      it's a more compatible kludge

0.3
   made it use .= throughout - I somehow had the idea that that was one
      of the things Perl has and PHP doesn't.
   made it set the image and title for a feed unconditionally, which
      makes for faster recovery if bad data should get stored for those.
      Thanks Les Benson for reporting the problem that led to this change.
   changes to purging algorithm to handle very slow feeds
   improved internal documentation
   added a time limit and a probabilistic update feature, both to improve
      response for systems that aggregate from a lot of different inputs

0.2
   added <HR> between entries
   made eliminate outdated entries (shame on me, I meant to do that in 0.1)

0.1
   initial release