In our project, if you run rake test, terrible things happen; you need to run rake spec. I can't seem to figure out how to re-define rake test to just output a message suggesting running rake spec instead.
How can I do this?
On your Rakefile at the end:
Rake::Task["test"].clear
task 'test' do
puts "use 'rake spec'"
end
Or even better
Rake::Task["test"].clear
task 'test' do
Rake::Task["spec"].invoke
end
Related
We wrote some tests that are necessary, but very slow. So we configured RSpec to exclude them except on Solano, where we set up an ENV variable.
# spec_helper
unless ENV['RUN_ALL_TESTS'] == 'true'
config.filter_run_excluding :slow
end
That works, but I'm trying to write a rake task we can call to run every test locally by setting that same ENV variable and then running the suite. I'm having trouble figuring out how to trigger RSpec. This is what I've got now:
# all_tests.rake
require 'rspec/core/rake_task'
desc 'Run all tests, even those usually excluded.'
task all_tests: :environment do
ENV['RUN_ALL_TESTS'] = 'true'
RSpec::Core::RakeTask.new(:spec)
end
When I run it, it doesn't run any tests.
Most of the stuff I found is for triggering a rake task inside of a test, testing a rake task, or setting up a Rakefile. We're using rspec-rails, our default rake task is already set up.
To run RSpec through its rake integration, you need to both define a task and invoke it:
# all_tests.rake
require 'rspec/core/rake_task'
# Define the "spec" task, at task load time rather than inside another task
RSpec::Core::RakeTask.new(:spec)
desc 'Run all tests, even those usually excluded.'
task all_tests: :environment do
ENV['RUN_ALL_TESTS'] = 'true'
Rake::Task['spec'].invoke
end
Rake::Task['spec'].invoke did nothing when you tried it because rake turns a task name which is not a name of a defined task but is a file name into a Rake::FileTask, both on the command line and in Rake::Task. You had no 'spec' task defined, but you have a spec directory, so rake spec ran without error and did nothing.
I had a similar problem where bundle exec rake default would properly create an RSpec::Core::RakeTask, but bundle exec rake spec would instead create a a Rake::FileTask on the spec directory, with with bundle exec rake spec --trace outputting:
** Invoke spec (first_time, not_needed)
It turned out that rspec-rails was in the :test gem group, where it needed (per the docs) to be in both :test and :development.
Interestingly (?), once I did that, gems that had previously only been in the :test group were also available from the specs even when launched with RAILS_ENV=development. I assume that rspec-rails engages in some magic environment shenanigans behind the scenes.
I'm trying to setup my rails project so that all the verification required by a contributor is in one command, currently we have been running:
rake test
But now we also want to use rubocop for static analysis:
rubocop -R -a
I want this to be executable in one simple rake task. It would be nice to override 'rake test' to run rubocop then the standard rake test stuff for a rails project, as then no-one will have to remember to change the command. But if I have to create a separate rake task, that's probably fine too.
I've seen the rubocop rake integration here, at the bottom, but I'm not sure how to bundle that with 'rake test' into one task... Any thoughts?
I prefer to set my default task to run rubocop then the tests. Either way, it is a good idea to have those tasks separate rather than have one task do two things.
require 'rubocop/rake_task'
task :default => [:rubocop, :test]
desc 'Run tests'
task(:test) do
# run your specs here
end
desc 'Run rubocop'
task :rubocop do
RuboCop::RakeTask.new
end
Your tasks:
> rake -T
rake rubocop # Run rubocop
rake test # Run tests
This is the .rake file that I ended up with in the end.
desc 'Run tests and rubocop'
task :validate do
Rake::Task['rubocop'].invoke
Rake::Task['test'].invoke
end
task :rubocop do
require 'rubocop'
cli = Rubocop::CLI.new
cli.run(%w(--rails --auto-correct))
end
You could easily define your own rake task which first invokes Rails' test rake task and then the code snippet you mentioned for rubocop.
For example, in a .rake file you could have something like that:
require 'rubocop/rake_task'
desc 'Run tests and rubocop'
task :my_test do
Rake::Task['test'].invoke
RuboCop::RakeTask.new
end
If you feel the need to customize the call to Rubocop and that involves more code, you could create another custom task, say :rubocop, which you then invoke from :my_test as well.
Finally, an alternative to creating your own rake task and sticking with rake test would be to modify your test_helper to invoke whatever you need invoked after testing is completed.
It appears to have changed from:
Rubocop::RakeTask.new
to:
RuboCop::RakeTask.new
See ma? I know how to CamelCase!
I used following .rake file to run the test and rubocop tasks at once:
task default: %w[rubocop test]
RuboCop::RakeTask.new(:rubocop) do |task|
task.patterns = ['**/*.rb']
task.fail_on_error = false
task.options = ["--auto-correct-all"]
end
task :test do
ruby 'test/program_test.rb'
end
The first line allows both tasks to be run by calling rake.
Command line arguments can also be added in the options array.
Here are the first few lines of my rake task:
desc "Import National Data into DB\n Usage:
rake import_regional[/Users/am/0925/xyz.csv, some_market]"
task :import_regional, [:file_path, :market] => [:environment] do |t, args|
.......
Here is how Im trying to run this task on the command line:
rake import_regional ["/Users/xyz.csv", "northeast"]
when I try this, I get this error
rake aborted!
Don't know how to build task
'import_regional[/Users/xyz.csv,'
I suspect it's the way in which you're running the task on the command-line; the [] doesn't have its Ruby meaning, there. Instead, maybe see if this works for you:
rake "import_regional[/Users/xyz.csv, northeast]"
This should still allow for spaces, etc. in :file_path, if required.
Peace,
tiredpixel
Command line isn't a Ruby interpreter, it's mangling things before it gets to Rake.
Try this:
rake "import_regional['/Users/xyz.csv', 'northeast']"
rake --tasks takes about 18s to run. This is just the time it takes to load all the tasks, as a result any task I define will take at least this amount of time to run :
$time rake --tasks
rake db:clean # Cleaning up database
rake passenger:restart # Restart Application
rake spec # Run specs
real 0m18.816s
user 0m7.306s
sys 0m5.665s
My Rakefile :
$: << "."
require "rubygems"
require "rspec/core/rake_task"
desc "Run those specs"
task :spec do
RSpec::Core::RakeTask.new(:spec) do |t|
t.rspec_opts = %w{--colour --format progress}
t.pattern = 'spec/*_spec.rb'
end
end
task :default => :spec
Any idea why rake takes to much times ?
Thanks
Try spring
Command line will look like:
spring rake -T
It will take more time running the first time, but subsequent runs will be very fast.
This solution worked for me: Faster rake tasks in Rails.
I had to do a little variation where I created a lib/tasks/no_rails directory and put all the Rake files which do not need Rails in there and loaded only those using the above method.
I like the solution Pratik mentions for the general case of loading rails for tasks that need it and not for those that don't, for any rake task without having to remember beforehand.
A less-invasive method to run a rake task that doesn't need rails is to use the -f rake option to tell rake to use a particular Rakefile. This way, rake won't go looking for rake tasks in all of rails.
For example, assuming your task above is in a file called Rakefile at the top level of your project and your Rakefile doesn't do anything that loads Rails like require File.expand_path('../config/application', __FILE__), you can do:
$ rake -f Rakefile spec
and it should run your spec task much faster. Try $ time rake -f Rakefile -T; I did this with a rails-independent Rakefile of mine and got:
real 0m1.543s
user 0m1.308s
sys 0m0.201s
The downside is you have to remember to specify this option every time, and not to specify it if you want to run a rake task from rails like rake db:migrate.
The entire rails environment has to be loaded, therefore even simple rake tasks such as rake --tasks take a while. Opening a console with rails console or script/console takes a similar time. You may try to hack Ruby or Rails to speed up rake, but too much optimization can be bad if you want to switch to a newer version later. Since the rails environment must be loaded, cleaning up routes may also help.
I would like to run db:migrate VERSION=0 and then db:migrate inside of my own rake task. I am confused about how to do this. Do I need a special require statement? My rake task will reside in the lib/tasks directory of a Rails app. Thanks.
Is your task just dependent on having a clean db? If that's the case then you can do:
task :my_task => [:environment, 'db:reset']
EDIT: Rake::Task[] won't accept parameters, you have to set it in ENV. In addition, you have to reenable the task to run it multiple times.
ENV['VERSION']= '0'
Rake::Task['db:migrate'].invoke
Rake::Task['db:migrate'].reenable
ENV.delete 'VERSION'
Rake::Task["db:migrate"].invoke
NOTE: Rake::Task.reenable requires Rake 0.8.2 or higher.
Check out rake db:reset as that will accomplish what you are trying to do.
To see what all of your rake tasks do, run rake -T