I have a private gem (built with jeweler) hosted on Github. I run unit tests before I commit but I'd like to test everything works when I use bundler from the end-user side.
So far I've just been creating a test environment with two files:
Gemfile:
gem 'my_gem', :git => 'git#github.com:my_repo/my_gem.git'
main.rb:
require "rubygems"
require "bundler/setup"
require "my_gem"
# Some code calling arbitrary methods from my gem
It seems like there's probably a way for me to run the unit tests built in to the gem.
Bundler supports the use of gemspecs for development. You can change your gemfile to the following:
source "http://www.rubygems.org"
gemspec
This will allow you to access all the gems you would need as a client of the gem you're building. Then in your test file you can just:
require "my_gem"
You can find more information in Yehuda's post.
Related
So, I've created a GitHub to manage my latest Ruby project, and I want for it to utilize a couple of gems. On my PC, all I have to go is type
gem install "gemName"
and it loads it to my computer, and then all I have to do in my Ruby script is have
require "rubygems"
require "gemName"
How can I do this with GitHub? What I tried to do is create a subfolder from the main repository (called "RubyGems") and then in my main ruby script
require "/RubyGems/colorize"
require "/Rubygems/psych"
With the two gems (colorize and psych) in the "RubyGems" folder.
Is this the proper way to do this? Will this even work? What is the right way to do this? (Sorry, I'm kinda new to GitHub.)
A couple of things, unless you're using a really old version of Ruby (like 1.9) you don't need to require 'rubygems' because is already required by default, next I highly recommend you to get familiar with bundler.
Bundler is used for "bundling" the required gems you use, to so do you have to install the gem (gem install bundler) and then you create a Gemfile, like this:
source 'https://rubygems.org'
ruby '2.2.0'
gem 'colorize', git: 'https://github.com/fazibear/colorize.git'
gem 'psych'
Execute bundle install after, that will create Gemfile.lock file, make sure you push both files to your repository.
With that you would be able to bundle exec ./your-script.rb, assuming your script is something like this:
require 'psych'
require 'colorize'
# Here I do stuff with psych and colorize
I have a test suite that is designed to run locally and on a cloud vendor. When the test is run locally I would like to use a RubyGem that makes use of native extensions that run much faster. On the cloud vendor however, I cannot include this particular flavour of gem and must use a different one.
Is there a way that I can, either by command line argument, or as a function in Ruby add or remove gem requirements?
You can wrap your require statements inside an if/else.
if `hostname`.strip == 'local'
require 'ruby_ext_gem'
else
require 'ruby_native_gem'
end
Or run the Ruby interpreter with the -r option.
ruby -rruby_ext_gem -rruby_ext_gem2 script.rb
Maybe bundler would be what you are searching for?
#Gemfile
source 'https://rubygems.org'
group 'local' do
gem 'ruby_ext_gem'
gem 'ruby-ldap', require: 'ldap'
end
group 'cloud' do
gem 'ruby_native_gem'
gem 'ruby-net-ldap', require: 'net/ldap'
end
# local
bundle install --without cloud
# cloud
bundle install --without local
#script.rb
#!/usr/bin/env ruby
require 'bundler/setup'
Bundler.require(`hostname`.strip)
I try to develop a gem. I include it in my rails application through Gemfile with :path option for testing purpose, but there are some errors that appear when I build it and release to rubygems that does not appear when gem included from local path. How can I install gem from *gem file (made with rake build command) in rails or any bundle driven application for testing?
This will give you the best help: http://guides.rubygems.org/make-your-own-gem/
But in summary you need to build your gem from the .gemspec then using irb require your gem to test it out.
Example:
gem build hola.gemspec
% irb
require 'hola'
=> true
I'm developing a gem with Jeweler in a custom directory.
I want to be able to require the gem within any app (and also the executables files from $PATH), and without needing to build and install the gem each time I modify it.
I thought about 2 ways:
I make a symlink to $GEM_HOME/gems and $GEM_HOME/bin
I add the bin directory to $PATH and the lib directory to rubygems to be loaded.
But I bet there is a proper way to do this.
You can specify a local path in the gem command:
gem 'your-gem', '1.2.3', :path => 'path/to/your-gem'
Update: As #Nick points out in the comments,
This is specific to using bundler. In general, it's just require '/path/to/your-gem.
I'd like to add, however, that if you're using a custom-developed gem, bundle will probably make your life easier if you're not already using it. This is because with bundler, when you're done developing the gem (or at a stable/release point) you can load a gem directly from a github repository like this:
gem 'your-gem', :git => 'git#github.com:you/your-gem.git'
No need to mess around with your Gemfile
require 'bundler/setup' will put the right gem into your $LOAD_PATH and allow you to require it on the next line.
#!/usr/bin/env ruby
require 'bundler/setup'
require '<gem-name>'
I have a project using cucumber outside of rails. How can I load the gems with the versions specified in my gemfile?
Digging through the Bundler website:
Create Gemfile (run bundle init to create skeleton Gemfile)
bundle install
In your app:
# Only needed for ruby 1.8.x
require 'rubygems'
# The part that activates bundler in your app
require 'bundler/setup'
# require your gems as usual
require 'some_gem'
# ...or require all the gems in one statement
Bundler.require
Could be worth checking out:
Bundler.io - Using Bundler in Your Appplication
Bundler.io - Bundler.setup and Bundler.require
Are bundle exec and require 'bundler/setup' equivalent?
I just learned about a way to make Bundler automatically require dependencies from a Gemfile. Add this code at the beginning of a Ruby program that has a Gemfile:
require 'rubygems'
require 'bundler/setup'
Bundler.require
With Bundler.require there's no need to explicitly require the gems/libraries enumerated in the Gemfile.
This solution is from http://technotales.wordpress.com/2010/08/22/bundler-without-rails/
To be honest I'm not sure if the require rubygems part is needed either.
Here's the simplest and most straightforward approach:
bundler init will create the Gemfile for you
Specify gems in the Gemfile.
Add the following to your main Ruby file
require 'bundler/setup'
Bundler.require
Run bundler install to install the gems.
More information can (now) be found at http://bundler.io.
Casper has a pretty good answer (despite some passive aggressiveness) but I think the missing piece for you is bundle exec. When you run the $ rails ... commands on the command line, Rails uses bundler to load those dependencies/gems. Rake, for example, doesn't by default so in order to run rake test using an older version of cucumber than what is on your system, you have to use bundle exec rake test. It's a good habit to get into always using $ bundle exec ... when you're using Bundler — it's explicit, you're always sure you're using the right gems, and it ensures you don't forget to add a dependency to your Gemfile (i.e. you push to another server or another developer and they are having issues because you didn't note the need for something you use but they don't).