Commits on Mar 2, 2012
  1. Add Jifty::Filter::JSON

    Luke Closs committed Mar 2, 2012
Commits on Feb 21, 2012
  1. @tsibley

    Bump version for external deps

    tsibley committed Feb 21, 2012
  2. @tsibley

    Append clickable query params _before_ any hash fragment

    Otherwise the query params become part of the fragment.
    tsibley committed Feb 21, 2012
Commits on Feb 13, 2012
  1. bump the version

    Luke Closs committed Feb 13, 2012
  2. Add a WebDriver layer, similar to Selenium

    Luke Closs committed Feb 13, 2012
Commits on Dec 5, 2011
  1. @tsibley
  2. @tsibley
Commits on Dec 3, 2011
  1. @alexmv

    Fix pod header sizes

    alexmv committed Dec 3, 2011
  2. @alexmv

    Allow a site_config.yml to override config.yml's plugin configuration…

    …, based on plugin class name
    alexmv committed Dec 3, 2011
  3. @alexmv
Commits on Nov 7, 2011
  1. @tsibley

    Bump version for external deps

    tsibley committed Nov 7, 2011
  2. @tsibley

    Preserve current request headers when returning from a continuation

    Request headers aren't pulled from env, they are a separate object.
    tsibley committed Nov 7, 2011
Commits on Nov 3, 2011
  1. @tsibley

    Selenium tests require an actual server

    Otherwise they hang pretty badly.
    tsibley committed Nov 3, 2011
  2. @tsibley

    Serialize actions to their moniker for JSON

    This is mostly to catch places where Jifty::Action objects are used
    instead of the moniker in a webservices request.  Without a method of
    serialization, blessed objects cause JSON to throw a fatal error.
    tsibley committed Nov 3, 2011
  3. @tsibley

    Request map action arguments submitted via YAML or JSON

    Webservices requests (aka ajax) fill out the Jifty::Request from a JSON
    or YAML data structure in the POST body instead of query parameters.
    Actions submitted via ajax can now take advantage of Jifty's request
    mapper for their arguments, i.e.
    
        my_action_arg => { result_of => 'moniker', name => 'result-field' }
    
    The hashref value is left as-is over the wire instead of encoded to a
    magical J:M-* parameter like non-ajax requests, and
    Jifty::Request::Mapper->map already has logic for dealing appropriately
    when 'source' is a hashref.  Values which are hashrefs but don't look
    like a request mapper spec are passed through untouched.  Specifically,
    Jifty::Request::Mapper->query_parameters is called by ->map and looks
    for one of the following hash keys to determine if the value should be
    mapped: request_argument, result, result_of, argument, argument_to.
    tsibley committed Oct 25, 2011
  4. @tsibley
Commits on Oct 24, 2011
  1. @obra
  2. @obra
Commits on Oct 18, 2011
  1. @tsibley

    Extract action order information more robustly when AJAX submitting

    Not all registration elements (such as those we generate in
    buttonFormElements) have ids, but all input elements will have names.
    If there's no id, check against the name attribute.
    tsibley committed Oct 18, 2011
Commits on Oct 12, 2011
  1. @tsibley
Commits on Sep 26, 2011
  1. @tsibley

    Don't specify an absolute path to firefox -- let Selenium search

    This reverts part of 59d12e7 which
    miscommitted what looks like local Selenium testing changes in the midst
    of a slew of completely different log changes.
    tsibley committed Sep 26, 2011
  2. @tsibley

    Don't call done_testing if the test is marked as no_plan

    Otherwise a plan is output twice, leading to soft test failure.  This
    happened with t/TestApp-JiftyJS/t/2-behaviour.t.
    tsibley committed Sep 26, 2011
  3. @tsibley

    Merge branch 'lukec/master'

    tsibley committed Sep 26, 2011
Commits on Sep 23, 2011
  1. @tsibley

    Print more than just the first argument in our overridden version of …

    …print
    
    Respecting $, and $\ (as described in `perldoc -f print`) probably isn't
    necessary, but it's more correct.
    tsibley committed Sep 23, 2011
  2. @tsibley

    Test that printing with multiple arguments works

    Even if it's deprecated, we should still print everything.  This adds a
    failing test.
    tsibley committed Sep 23, 2011
Commits on Sep 22, 2011
  1. Finish testing so no_plan tests are happy.

    Luke Closs committed Sep 22, 2011
Commits on Sep 9, 2011
  1. Support using a subclassed Test::WWW::Selenium

    Luke Closs committed Sep 9, 2011
Commits on Aug 10, 2011
  1. @obra

    Test::WWW::Mechanize no longer treats content_lacks as content_unlike,

    breaking the existing tests
    obra committed Aug 10, 2011
Commits on Aug 4, 2011
  1. @alexmv

    Checking for path hygene in continuation returns is no longer necessary

    This warning was introduced in 36bd338, when "call" and "return" were
    handled by the same code.  The only way we determined which phase we
    were in was by whether the path matched what was stored in the
    continuation.  The danger is this was that the path that we redirected
    to could be different from the path that the browser then requested, if
    it chose to escape characters differently; this would cause Jifty to
    redirect indefinitely.  The warning was placed here to warn of the
    perils of redirecting to URIs containing characters which the browser
    might choose to escape differently than Jifty did.
    
    However, the canonicalization done by Plack::Request (and in 3e54cf2)
    guarantees that the URIs have been canonicalized equivilently; thus,
    this warning is always a false-positive.
    alexmv committed Aug 4, 2011
Commits on Aug 3, 2011
  1. @tsibley

    Jifty.update supports before/after region insertion modes, so form el…

    …ements should too
    
    While we're at it, clarify the documentation around append/prepend and
    before/after.
    tsibley committed Aug 3, 2011
Commits on Jul 15, 2011
  1. @tsibley

    Force the default configured URL when called from a Jifty::Notification

    This means we respect your canonical hostname and scheme instead of
    basing it on the irrelevant current request that caused the email
    generation.
    
    Cloning fixes a bug where we forced scheme to http but didn't touch
    port, potentially leaving it as 443 and generating URLs like
    http://example.com:443/.
    tsibley committed Jul 15, 2011
Commits on Jul 13, 2011
  1. @alexmv
Commits on Jul 12, 2011
  1. @tsibley

    Try to be less brittle if we find ourselves without a placeholderText…

    … attribute
    
    Behaviour will still be wonky, but javascript errors won't abound.
    tsibley committed Jul 12, 2011
  2. @tsibley

    Only reinstate a placeholder if the input had one before we cleared it

    Otherwise empty inputs are blindly marked as having placeholders when
    they never did.  This causes some hilarious behaviour when
    ajax validation/canonicalization runs and leads to javascript errors.
    tsibley committed Jul 12, 2011
  3. @tsibley

    Template args should be able to be 0

    set(foo => 0) works, but you can't pull it back out with get('foo')
    unless you test definedness, not truth.
    
    We really want the // operator here.  Hopefully this doesn't break any
    assumptions about get(); the tests do pass.
    tsibley committed Jul 12, 2011