1.7 2008-08-28
  FEATURES
   - added interactive docs plots to sweb (see sweb/help.html#Plots)
   - plot with gnuplot instead of R (reduces install and run overhead)
   - added ability to capture relevance feedback for viewed messages
   - added bruteforce but customizable message templating
   - factor to color terms by can now be set via SWEB_TERMS_SORT
   - added security filter so users can only see their allowed hosts
  FIXES
   - sstats now correctly handles matricies with indexes > ULONG_MAX
   - dependencies reduced, and better checked by configure
   - sloghound made more robust

1.6 2008-04-14
  FEATURES
   - term query output now includes count per nodehour (handy access
     to time and node rate of occurrence; segrep)
   - rows of matching stats are visually grouped (indicating
     co-occurring terms or templates; sweb)
   - table headers no longer scroll with table data (sweb)
   - consolidated term configurations, adding the ability to specify
     "alert" terms (which can be queried for in sweb with a single
     click; termconf)
   - improved memory handling, enabling analysis of larger log sets (sstats)
  FIXES
   - more robust importing of large log sets (syslogs2docs)

1.5 2007-09-10
  FEATURES
   - drastically reduced compute time and memory requirements by
     replacing R calculations with a standalone C program (sstats)
 
1.4 2007-08-13
  FEATURES
   - drastically increased word-based query performance and capability
     via index-based search including regexps
   - increased time-based query performance
   - added many convenience links in terms frame
   - decreased memory footprint of R calculations
   - added autoconf and Debian packaging
   - improved man pages and added examples directory
   - improved command line useability
  FIXES
   - reduced liklihood of sweb errors due to incomplete stats files
   - updated usage lines and man pages

1.3  2007-02-19
  FIXES
   - sweb sort didn't obey scientific notation (it has now been
     eliminated because it keeps the underlying unix sort faster, and
     seems good for human readability too)
   - vct used to give empty output if PROG_PID_POS was not set in
     sisyphus.conf (now it has a default value of 1)
   - index_merger used to require -s to work properly (now it has a
     default value of 2)
   - sprep -months used to not work correctly
  FEATURES
   - memory and cpu requirements are significantly reduced during
     factor aggregations (eg during host and time information
     calculations)
   - splot(url=sweb) will use firefox instead of xterm to display
     colorized logs (see "?splot" at the R prompt for more info)
   - syslogs2docs -groups will divide logs into separate SISYPHUS_DATA
     groups, to enable grouping by similar-function-hosts
   - syslogs2docs -incr can be used to incrementally import data
   - added documentation on how to ignore terms via index_merger -I
   - sprep stderr/stdout cleaned up for better use in cron

1.2  2006-11-28
  FIXES
   - fixed bug which caused the sweb message templates frame to not work
     (genrex was not being installed properly)
   - fixed bug which caused the term support CDF plot in
     SummaryPlots.pdf to be incorrect (corpus.terms-{unsup,ignored}
     are now generated correctly)
   - fixed bug which caused sweb terms frame to not work if terms longer 
     than ~400 chars are present (changed SparseInfo/R/write.terms.R)
  FEATURES
   - added functionality to ignore words based on regular expressions
     (can prune from the index using index_merger -I)

1.1  2006-11-11
   - addition of web interface (sweb) and MANY bug fixes and
     functionality improvements

1.0.1	patch from nathan dauchy (muchas gracias!):
   - added capability merge top n regexps in genrex 
   - slctify can be given more than one file, and prunes out log lines
     with empty msg section
   - added more info to INSTALL changed syslogs2docs to not die due to
     too many open files

1.0	added term/doc information stuff (indexer, SparseM, etc...)
0.9	initial release, message template capabilities only