The master branch of agility is set up to test Hobo 1.0.X with Ruby
1.8.7, Rails 2.2.2, sqlite3 and Firefox on Linux. The
directory may be used to modify any of these assumptions. For
example, if you wish to test Rails 2.3.11 and Hobo 1.1, type
# patch -p1 --merge < patches/rails-2.3.11.patch # patch -p1 --merge < patches/hobo-1.1.patch
Currently the only meaningful tests here are Webrat (aka integration) and Selenium (aka acceptance) tests. The unit & functional tests run, but don't contain any useful tests -- the unit tests for Hobo are in Hobo itself.
Integration tests should run with a simple
after you've done the standard
bundle install and
You may also need to do
rake hobo:generate_taglibs if you've never
run agility in the development environment.
To run the tests under mysql, copy
config/database.yml.mysql. DO NOT TOUCH
directly. Instead run
# patch -p1 --merge < patches/mysql.patch
Acceptance tests require a running server and a working browser, so
are more difficult to run. To make this easier, we've included an
acceptance_test.sh that you can run and/or examine. You may have to
config/selenium.yml if the test has trouble contacting your
Note that you'll probably want to tweak the hobo definition in
Gemfile and then run
bundle update hobo to test your branch of
Hobo rather than the stock version, which has already been tested.
Testing on multiple versions of Ruby and Rails
We regularly test Hobo against many versions of Ruby and Rails. All
the combinations that are tested are listed in
you want to do the same, here's some instructions.
The first thing you should do is to trim down
that it doesn't take several hours to run. Once you have things
figured out you can add more combinations into
The other first thing you should do is to make sure that
rake test:integration and
./acceptance_test.sh work for every database
you're testing against.
Install RVM & the Rubies
To make testing different versions of Ruby and different sets of gems easy, this environment uses the awesome tool rvm.
Install rvm as per instructions, and then install the rubies. You'll probably want to install some libraries and dependencies before the rubies.
So it might go something like this after rvm is installed:
rvm notes # cut and paste from the above output. On Ubuntu this would be: aptitude install curl sun-java6-bin sun-java6-jre sun-java6-jdk aptitude install build-essential bison openssl libreadline5 libreadline5-dev curl git zlib1g zlib1g-dev libssl-dev libsqlite3-0 libsqlite3-dev sqlite3 libxml2-dev rvm package install readline rvm package install openssl rvm package install iconv rvm package install zlib # this is only 2 of the rubies listed in test-combos.txt, but it's enough to start rvm install 1.8.7-p249 rvm install 1.9.2-p136 -C --with-readline-dir=$rvm_path/usr
You also have to have the mysql server installed for some of the combos, and you need to be able to build the mysql gem. For example, on Ubuntu, you should probably have these packages installed: build-essential ruby-full libsqlite3-dev mysql-server mysql-client libmysql-ruby libmysqlclient-dev libxslt-dev libxml2-dev libreadline-dev.
Create the gemsets
For test purposes we install our gems into a separate gemset for each ruby so we do not pollute your global gemsets. This gemset will be called hobotest.
Create all the gemsets and let bundler populate them
rvm use 1.8.7 # just making sure rvm is initialized ./rvm-create-gemsets.sh
This will fail if your git is not clean.
You can rerun
rvm-create-gemsets.sh whenever you've updated any
If the only dependency that's been updated is hobo, you can run
rvm-update-hobo.sh, which runs quicker.
Run the tests
You can filter the tests, which just does a grep in test-combos.txt. A few useful examples:
./test-all-combos.sh hobo-1.0 ./test-all-combos.sh jruby # tests both jruby's ./test-all-combos.sh jruby-1.5.1 ./test-all-combos.sh 1.8.7 ./test-all-combos.sh ruby-1.8.7-p249,rails-2.2.2,sqlite3,hobo-1.0
If you test with IE6/7, the first test "Autocomplete and contributors" will fail. This is a problem with the test, not Hobo. The second test "Editors" is flaky. In both cases the problem is properly waiting for actions to complete before continuing the test.
On other browsers, the first two Selenium tests will occasionally fail. These are timing errors. This happens infrequently enough that increasing the margins isn't worth the slowdown. If you are experiencing consistent failures on any test, please report it.