Permalink
Commits on Jul 3, 2016
  1. Skip t/36-email.t

    doherty committed Jul 2, 2016
    Need to figure out a new testing strategy here. [[User:Perlwikibot testing]]
    was created with a confirmed email so you could send emails to it. The
    account was then locked (in CentralAuth), but this still permitted emails
    to be sent. MediaWiki no longer allows this. We need to figure out another
    plan.
  2. Add a subtest for checking the page history for a successful edit

    doherty committed Jul 2, 2016
    This makes it easier to skip_all when the previous test didn't run
    (typically because of encountering a captcha)
  3. [travis] upload to S3 instead of p.hashbang.ca

    doherty committed Nov 14, 2015
    This requires additional secret config on travis-ci.org
    for the S3 credentials, which I've set up.
  4. [travis] Run tests on the new containerized infra

    doherty committed Nov 14, 2015
    This requires not using sudo; see
    http://docs.travis-ci.com/user/migrating-from-legacy/
    
    Although you can't use sudo to install packages during
    before_install, there is an APT add-on that allows you
    do do that: http://docs.travis-ci.com/user/apt/
    
    Also, build both dev and master
Commits on Feb 27, 2016
  1. Merge pull request #79 from stemd/master

    doherty committed Feb 27, 2016
    fixing test 29
  2. fixing test 29

    stemd committed Feb 27, 2016
    Hash returned now has one element more, it seemed more reasonable
    to add new element to expected hash in test than to eliminate that
    element from result of function just to pass the test.
    
    Hopefully my logic is sound. ;)
  3. Merge pull request #78 from stemd/master

    doherty committed Feb 27, 2016
    fix test t/25-sitematrix.t
  4. fix test t/25-sitematrix.t

    Darko committed Feb 27, 2016
    Simple stripping & prepending of URI prefix where needed for test to
    pass, 2nd possibility was to change test itself, but I like to keep
    backwards compatibility as much as possible. :)
Commits on Apr 18, 2015
  1. cleanup formatting

    doherty committed Apr 18, 2015
  2. Fix t/18-is_blocked.t

    doherty committed Apr 18, 2015
    The account we checked was renamed for SUL.
  3. Merge branch 'dev'

    doherty committed Apr 18, 2015
    Conflicts:
    	Changes
Commits on Jan 25, 2015
  1. Fix t/48-get_image.t

    doherty committed Jan 25, 2015
    Someone had changed the file to be much higher quality. Instead
    of continuting to use it, pick a smaller file. I've protected
    [[File::Wiki.png]] with a note explaining why.
Commits on Sep 27, 2014
  1. Documentation tweaks

    doherty committed Sep 27, 2014
Commits on Sep 9, 2014
  1. Released 5.006002

    doherty committed Sep 9, 2014
  2. Released 5.006001

    doherty committed Sep 9, 2014
  3. POD fixes

    doherty committed Sep 9, 2014
Commits on Sep 7, 2014
  1. Link to API documentation in POD

    doherty committed Sep 7, 2014
    Fixes gh-63
  2. Use HTTPS by default and warn if logging in over HTTP

    doherty committed Sep 7, 2014
    Also, update all tests that log in to use HTTPS.
  3. Remove the ability to set the wiki to test against

    doherty committed Sep 7, 2014
    Nobody uses this, and if you ever did it'd almost certiainly cause
    the test suite to fail unless you copy the test corpus from testwiki.
  4. Mark flaky tests as extended

    doherty committed Sep 7, 2014
Commits on Sep 3, 2014
Commits on Sep 2, 2014
  1. Released 5.006000

    doherty committed Sep 2, 2014
Commits on Sep 1, 2014
Commits on Aug 22, 2014
  1. Don't edit the operator's talk page if edit encounters errors

    doherty committed Aug 22, 2014
    This is bad because
      - this is fragile -- the edit routines don't do proper error handling
      - this is annoying
        - this might encourage operators to omit the operator parameter
          to the constructor to avoid getting these notices
      - this removes control of error handling from the developer
      - this only deals with errors encountered while editing
    
    Just remove it and let devs do the right thing.
    
    Fixes gh-68
  2. Let devs solve CAPTCHAs if needed

    doherty committed Aug 22, 2014
    edit will now return false but use a specific error code, ERR_CAPTCHA,
    when a CAPTCHA has been encountered. The error details begin with
    'captcharequired', and a special 'captcha' key is added with data
    regarding the CAPTCHA which must be solved.
    
    After interaction with a human, the edit can be retried, adding
    the new captcha_id and captcha_solution parameters (in the
    new hashref call style).
    
    Fixes gh-2
Commits on Aug 20, 2014
  1. Set some common options for prove

    doherty committed Aug 3, 2014
    In particular, specify tests which shouldn't run in parallel. The
    tests identified as "seq" will always be run sequentially, while
    all others are eligible for parallel execution. Specify -jN to
    run N test jobs.
  2. Catch when a CAPTCHA has been encountered

    doherty committed Aug 2, 2014
    This doesn't give you any way to deal with the CAPTCHA yet, but
    it can let you know what happened. Since CAPTCHAs aren't a "real"
    error condition, they're presented differently, and we didn't
    understand that, leaving no error information in the Bot object's
    error attribute. Now, we add a specific error constant and provide
    the data needed to proceed in the details. The next step is to
    augment the edit method with a way to accept both the ID and the
    solution, and retry properly. It might also be interesting to try
    an automatic interactive retry option.