Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
branch: master
Commits on Jan 1, 2015
  1. Better ids for tweets

    Because of not using my brain, I did a poor job of picking an id, the current
    year, really?  At least the bug showed up quickly.
Commits on Dec 23, 2014
  1. Add OpenBSD_sets twitter feed

    authored committed
  2. Slightly better disabling of saving new keys

    authored committed
    Don't remember why I had them disabled, I'm sure there was a good reason.
Commits on Mar 30, 2014
  1. Don't show the import conflict message in the tweet

    It doesn't improve it at all.
  2. Only combine spaces or tabs, not newlines

    Because it should make for a prettier tweet.  However, leaves open
    characters being eaten by having spaces at the end of a line.
  3. Refactor parse_log_message

    It was getting too much inside the main sub.
  4. Notice that the import details are inside of the Log Message

    So properly detect and handle them.
Commits on Jan 22, 2014
  1. Enable SSL to twitter

    It is required, and for some reason not the default.  Wish I had realized that
Commits on Aug 29, 2013
  1. Tweet changes to stable on OpenBSD_stable

    Because it is probably a good idea.
Commits on Aug 24, 2013
  1. Add the "tag" if there is one, indicates a commit to -stable

    Which is fairly useful.  Now that I know about it, I should setup a new
    feed, OpenBSD_CVS_Sec or somesuch.
  2. Make the logic of when to add a "wildcard" easier to read

    Too many negatives, unless not, ouch!
  3. Mention what we retweet

  4. Better decode the ChangeLog

    It's apparently in UTF-8 and we want to be able to show names like
    Jean-Gérard Pailloncy properly.
  5. Parse the ChangeLog instead of a Maildir

    Not huge changes, but important, means it is less likely to blow up if I don't
    clean out my mailbox and use less memory.
    Also includes some improvements in the parsing of the commit message, so
    should be fewer "did something the parser couldn't recognize" messages.
    Plus a script to sync the changelog
Commits on Jun 14, 2013
  1. Switch to twitter's v1.1 API

    Fortunately someone else did the work.
  2. Kill uninit warning

    that was annoying.
Commits on Jun 10, 2013
  1. Re-shorten tweets that fail because they are too long

    Stupidly I don't have a good way to figure out what twitter will accept and I
    can't ask twitter "truncate for me" so we just have to try and fail and try
Commits on Sep 29, 2012
  1. Die if we can't send the tweet

    That way we can sleep for a while and then come back and re-attempt sending
    any messages we missed.
  2. Sort the email we find by mtime before sending

    This should mostly send them in the right order, way more easily than sorting
    by time in one of the email headers.
Commits on Jul 9, 2012
  1. ReTweet other accounts on OpenBSD_cvs instead of real content

    It means we get all the tweets, but better icons and less duplication if you
    follow multiple.
  2. Convert get_seen() to just seen()

    which means don't tie a new handle every time it is called.
  3. Anchor the match so we don't get quoted commits

    Unlikely to happen because we should fail with the attribution line but if
    somehow it wasn't there it could cause issues.
  4. Try two for profiles

    They at least each have their own look so you can easily tell them apart
  5. Handle Imports

    They happen fairly frequently on ports so we want to show useful things.
  6. Friendlier wording when we don't know what was changed

    We at least know how much.
  7. Handle partial word chops better

    Somewhat better message if the last thing we removed wasn't a trailing /
  8. Better messages for files with regress tests

    Not quite right still for Imports but regress is better
Commits on Jul 8, 2012
  1. Split handling of Log Message out

    Makes the code simpler and now I can tell the sometimes the M is upper and
    sometimes lower case.
  2. Match Update commit messages

    Not sure if it is quite right but should be better.
  3. Just match on files, easier that way

    Probably still not ideal, but better and I'll need to put some thought into
Something went wrong with that request. Please try again.