Permalink
Commits on Jul 19, 2012
  1. Use public API of MiniTest for registering assertions.

    As suggested by @tenderlove in [1]. This reduces the dependency on
    private & potentially more changeable behaviour in MiniTest and gets us
    closer to the goal of not having to monkey-patch.
    
    [1] freerange#87 (comment)
    floehopper committed Jul 14, 2012
  2. Introduce a factory for expectation errors.

    The idea is that we'll be able to use this to raise exceptions for the
    appropriate test framework which should simplify the integration code
    and eventually hopefully mean we don't need to monkey-patch Test::Unit
    or MiniTest.
    floehopper committed Jul 13, 2012
  3. Don't fail so fast if no test library is loaded before Mocha.

    Fixes #88.
    
    This partially reverts c8eba68, by
    displays a deprecation warning instead of raising an exception.
    
    I hadn't realised that some test library integration still relied on
    requiring "mocha" rather than "mocha_standalone", so it seems kinder to
    give them some warning.
    
    I still intend to make this fail fast in a later release, because it
    resolves one of the biggest problems I see with people using Mocha.
    floehopper committed Jul 19, 2012
Commits on Jul 14, 2012
  1. Remove deprecated access to `Mocha::Standalone`.

    The use of `Mocha::Standalone` has been deprecated for years and should
    be safe to remove. Use `Mocha::API` instead.
    floehopper committed Jul 14, 2012
  2. Remove the deprecated file `stubba.rb`.

    This has been deprecated for years and should be safe to remove.
    floehopper committed Jul 14, 2012
  3. No need to specify this explicit version of `hoe`.

    Now we're using Rubygems >= v1.8.11 and Ruby >= v1.8.7, this
    doesn't seem to be a problem.
    floehopper committed Jul 14, 2012
Commits on Jul 13, 2012
  1. MiniTest v1.4.1 is broken on platforms not supporting SIGINFO.

    We get the following errors when running on Travis CI [1] :-
    
        ArgumentError: unsupported signal SIGINFO
    
    Note that I've kept the relevant Gemfile and still run the tests locally
    in the build-matrix script, because it works on OSX.
    
    [1] http://travis-ci.org/#!/freerange/mocha/jobs/1858953
    floehopper committed Jul 13, 2012
  2. Include minitest v1.4.1 in Travis CI build.

    This version has a distinct monkey-patch and we already had a Gemfile
    for it - must have just missed it out when writing the `.travis.yml`
    file.
    floehopper committed Jul 13, 2012
  3. Run build matrix with multiple versions of Ruby.

    Also the latest version of test-unit doesn't work with either version of
    Ruby.
    floehopper committed Jul 13, 2012
  4. A hacky way to run all the builds in the travis "matrix".

    Not currently loading `.travis.yml`.
    
    Found I had to additionally ignore `Gemfile.minitest.1.4.1` and
    `Gemfile.test-unit.latest`. Not sure why I had to do the former, but the
    latter is because a new version of `test-unit` is out which is currently
    not supported by `mocha`.
    floehopper committed Jul 13, 2012
Commits on Jul 5, 2012
Commits on Jul 4, 2012
  1. Attempt to get notifications sent to all @freerange peeps.

    The default is apparently to send email to the author & committer and
    all public members of the Github organisation, which should be what we
    want.
    floehopper committed Jul 4, 2012
  2. Rename method to make it less confusing.

    I think this is a carry over from an old name for the library.
    floehopper committed Jul 4, 2012
  3. No need to check `defined?` for outer modules.

    As suggested in [1].
    
    [1] freerange#40 (comment)
    floehopper committed Jul 4, 2012
  4. Do not monkey-patch Test::Unit or MiniTest unless we *know* it's ok.

    I had assumed both libraries were changing relatively slowly and that
    the Mocha Travis CI build in conjunction with the
    Gemfile.minitest.latest and Gemfile.test-unit.latest gemfiles would
    catch any problems, but this seems to have been overly optimistic.
    
    Also the logic in this set of if statements was flawed - the else
    statement was unlikely to ever be reached.
    floehopper committed Jul 4, 2012
Commits on Jul 3, 2012
Commits on May 14, 2012
Commits on May 4, 2012