Much of the publishing process is automated:

Write:
  * Develop stories in "content.hs". The file is an instance of the HWN
    data type, defined in utils/publish.hs
  * urls are in [http://a.b.c/ text] wiki format

  * Don't forget to get the quotes from lambdabot

Typeset:
  * When done writing the content, preview the issues with "./publish"
    which dumps rendered .txt, .html and .wiki files in the top level dir.

  * run 'linkchecker yyyymmdd.html',     and recurse

  * Once all is good, run "utils/publish -p", which will
    move the .html and .txt versions into the archives/ versions
    (suitable for email and sequence.complete.org), and also the wiki
    versions (Old_news.html and News.html). 

    This'll run the spell checker, which requires aspell installed. The
    issue number will be extracted (and then incremeneted) 
    from 'issue' (with -p)

  * The date comes from the current date.

  * darcs add
  * darcs record

Publish:
  * email archives/yyyymmdd.txt to haskell@haskell.org
  * upload archives/yyyymmdd.html to sequence.complete.org
  * upload wiki/News.html to haskellwiki/News
  * upload wiki/Old_news.html to haskellwiki/Old_news
  * flush haskell wiki cache
  * email summary and url to lwn@lwn.net:

        The October 31, 2006 edition of the 
        <a href="http://sequence.complete.org/hwn/20061031">Haskell Weekly News</a>
        is online. This week we see a number of community documentation
        and maintenance efforts, and the appearance of indexed data
        types in GHC.

  * post url to programming.reddit.com, suggested title:
         "Haskell Weekly News: headlines"

  * darcs tag 

Record:
  * darcs record