How to remove another gem with my gemspec - ruby

We have a gem foo that used to have a dependency on another of our gems bar. But now we've pulled the bar code directly into foo.
Is it possible to have the bar gem removed the next time our users upgrade foo?

I seem to have found a quick and dirty solution, similar to adding conditional install time deps. Unfortunately, I haven't figured out how to tell the user, since gem squashes the extension output unless the user specifies verbose.
in the ext/mkrf_conf.rb, I added:
require 'rubygems'
require 'rubygems/uninstaller.rb'
begin
gem = Gem::Specification.find_by_name('other_gem');
Gem::Uninstaller.new('other_gem').uninstall
rescue Gem::LoadError
end

Related

Two gems share same require?

When I call:
require 'retryable'
These two gems clash:
https://github.com/robertsosinski/retryable
https://github.com/carlo/retryable
as they both have a 'retryable' file they ask the user to require. I'm interested in using the first gem, however this doesn't always happen.
This code is executed as a part of my own gem, and it has to be reliable across all users.
Is there a way to require specifically from a gem (as the gem names are different of course)?
How do I resolve this naming conflict?
EDIT: To clarify, this is the official repo and the gem names are actually different ("retryable-rb" and "carlo-retryable"), however they both ask their users to require the lib/retryable.rb file with require 'retryable'
You can explicitly activate a specific gem with the gem method.
In this case you want the retryable-rb gem, and not any others that may have a retryable.rb file:
gem 'retryable-rb' # activates the gem in question
# and adds its lib dir to load path
require 'retryable' # loads retryable.rb from the retryable-rb gem, as it
# is now on the load path

Pry is not a module

Ok, as #tim-moore asked, I will post it in new question.
Ok, so I wanted to make gem using bundle. Pry extension gem require that gem start with pry- as mentioned here.
I used:
bundle gem pry-name
but it messed up my file structure
create pry-name/pry-name.gemspec
create pry-name/lib/pry/name.rb
create pry-name/lib/pry/name/version.rb
As you can see it created lib/pry directory. I know it's gem's style to created such structure but now I pry cannot load this gem automatically
One solution from my question was:
create pry-name.rb that contain only require 'pry/name'
After I have done this, and build gem, I started pry:
This message appear:
require 'pry-name' # Failed, saying: Pry is not a module
As for my guesses:
I'm creating commands writing something like this:
Pry::Commands.create_command "name-of-command" do
# my code goes here
end
and, as ruby find Pry::Commands. it want require it from lib directory not from Pry gem.
What does this error mean. Why it doesn't work. How make it work keeping in mind gem and pry requirements(pry gem starts with pry- and gem will create another directory(ies) when someone use - for example: gem pry-name will make pry/name)
Everywhere in your newly-created gem where it has module Pry, change it to: class Pry. Since Pry is already defined (as a class), you cannot redefine/reopen it as a module.

How to add "pry" when developing a Ruby gem

I have a gem called "something".
I would like to add pry as a development dependency when developing the gem. However I don't know how to load it.
If I have "require something" inside lib/something.rb , when I release the gem, it throws a LoadError, because pry is only a development dependency.
At the same time I don't want to keep adding and removing pry when I am committing code.
What is the best way to require pry only when developing the application, but not require it as a dependency for the gem?
You can use the add_development_dependency in the gemspec file. You'll still have to require it in your lib/something.rb file within a begin .. rescue LoadError block. (Edit 2, see below)
In your case, it will be something like the following:
spec.add_development_dependency 'pry', '~> 0.9.12.2'
The purpose of add_development_dependency is to separate the gems into dependencies that get installed when you execute gem install mygem vs development-only dependencies that are installed only when you execute gem install mygem --development.
Edit: #Pierre-Louis Gottfrois' solution
Modify the Gemfile directly and add a test group. This question describes the process. This does not appear to be a preferred solution according to Yehuda Katz.
Edit 2: begin require ... rescue LoadError is apparently a common practice for Ruby scripts, according to this Making Ruby Gems article.
I think I found a workaround for that.
If you configure bundler to use pry as your console with
$ bundle config console pry
Then pry is itself required and you don't need to explicitly require in your source files.
Plus, you get a history on pressing ' ↑ '.

Is it possible to require files outside the Gemfile?

For example, I'm developing a gem, and while I'm developing, I use pry instead of IRB, and debugger for debugging. However, I don't want possible contributors to have to install them (because they may not need them). My first idea was to put them in a Bundler group:
source :rubygems
gemspec
group :extras do
gem "pry"
gem "debugger"
end
And then people could use:
$ bundle install --without extras
But I want it to be a default that they're not installed. What would be perfect is that they're not in my Gemfile, but that I can still require them (if they exist on the computer). This solution would be ok because I don't care at which version they're locked. Can it be done?
You can add arbitrary load paths and then require gems from them. Check out the global variable $:
puts $:.inspect
# ["/var/myproject/releases/20200918191637/lib", "/var/myproject/releases/20200918191637/vendor", "/var/myproject/releases/20200918191637/app/assets", "/var/myproject/releases/20200918191637/app/controllers", "/var/myproject/releases/20200918191637/app/helpers", "/var/myproject/releases/20200918191637/app/mailers", "/var/myproject/releases/20200918191637/app/models", "/home/deploy/.rvm/gems/ruby-2.3.0/gems/bundler-2.0.2/lib", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/gems/2.3.0/gems/json-1.8.3/lib", "/home/deploy/.rvm/gems/ruby-2.3.0/gems/bundler-2.0.2/lib/gems/bundler-2.0.2/lib", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/site_ruby/2.3.0", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/site_ruby/2.3.0/x86_64-linux", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/site_ruby", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/vendor_ruby/2.3.0", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/vendor_ruby/2.3.0/x86_64-linux", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/vendor_ruby", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/2.3.0", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/2.3.0/x86_64-linux"]
Now let's append to $: and require a gem
require 'method_source'
# LoadError: cannot load such file -- method_source
$: << '/home/deploy/.rvm/gems/ruby-2.2.4#myset/gems/method_source-1.0.0/lib'
# ["/var/myproject/releases/20200918191637/lib", "/var/myproject/releases/20200918191637/vendor", "/var/myproject/releases/20200918191637/app/assets", "/var/myproject/releases/20200918191637/app/controllers", "/var/myproject/releases/20200918191637/app/helpers", "/var/myproject/releases/20200918191637/app/mailers", "/var/myproject/releases/20200918191637/app/models", "/home/deploy/.rvm/gems/ruby-2.3.0/gems/bundler-2.0.2/lib", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/gems/2.3.0/gems/json-1.8.3/lib", "/home/deploy/.rvm/gems/ruby-2.3.0/gems/bundler-2.0.2/lib/gems/bundler-2.0.2/lib", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/site_ruby/2.3.0", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/site_ruby/2.3.0/x86_64-linux", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/site_ruby", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/vendor_ruby/2.3.0", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/vendor_ruby/2.3.0/x86_64-linux", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/vendor_ruby", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/2.3.0", "/usr/local/rvm/rubies/ruby-2.3.0/lib/ruby/2.3.0/x86_64-linux", "/home/deploy/.rvm/gems/ruby-2.2.4#myset/gems/method_source-1.0.0/lib"]
require 'method_source'
# true
That can be tedious if you have a lot of dependencies, though, so you might try this:
Dir['/home/deploy/.rvm/gems/ruby-2.2.4#myset/gems/*'].each do |gem_path|
$: << File.join(gem_path, 'lib')
end
bundle install is "opt-out"—unless you specify --without some_group, it installs everything.
If you absolutely don't want to have a given gem in your Gemfile, you could just gem install that rogue gem outside of your bundle. Then it'll be visible to you under irb and straight ruby (but obviously you'll get errors if you try to require it within code running under bundle exec).
You could add a conditional based on environment variables into the Gemfile. Example:
source :rubygems
gemspec
if ENV['WITH_EXTRAS'] == '1'
gem "pry"
gem "debugger"
end
The gems are then only installed/loaded, if you set the environment variable to '1' e.g. WITH_EXTRAS=1 bundle install.
Nowadays you can use the "optional" argument, and it will do exactly what you ask: it will not install the extras unless people bundle --with extras
group :extras, optional:true do
...
end
https://bundler.io/v2.3/guides/groups.html#optional-groups

Check for Ruby Gem availability

Is there a way to check if some gem is currently installed, via the Gem module? From ruby code, not by executing 'gem list'...
To clarify - I don't want to load the library. I just want to check if it's available, so all the rescue LoadError solutions don't help me. Also I don't care if the gem itself will work or not, only whether it's installed.
In Ruby 1.9.3 only there is also:
Gem.available?('somegem')
You can use regex expressions too. Handy if I want to allow 'rcov' and GitHub variants like 'relevance-rcov':
Gem.available?(/-?rcov$/)
Looking at the Gem API documentation, using Gem::Specification::find_all_by_name to test for gem availability seems reasonable.
if Gem::Specification::find_all_by_name('gemname').any?
do stuff
end
find_all_by_name always returns an array (of Specification objects), as opposed to find_by_name which raises an exception if no match is found.
IMHO the best way is to try to load/require the GEM and rescue the Exception, as Ray has already shown. It's safe to rescue the LoadError exception because it's not raised by the GEM itself but it's the standard behavior of the require command.
You can also use the gem command instead.
begin
gem "somegem"
# with requirements
gem "somegem", ">=2.0"
rescue Gem::LoadError
# not installed
end
The gem command has the same behavior of the require command, with some slight differences. AFAIK, it still tries to autoload the main GEM file.
Digging into the rubygems.rb file (line 310) I found the following execution
matches = Gem.source_index.find_name(gem.name, gem.version_requirements)
report_activate_error(gem) if matches.empty?
It can provide you some hints about how to make a dirty check without actually loading the library.
Since Gem.available? is deprecated (argh!), you have to rescue again (double aaargh). Yes, find_by_name throws an exception if the gem is not found. So to be backwards-compatible with older rubygems, the common solution seems to be :
def gem_available?(name)
Gem::Specification.find_by_name(name)
rescue Gem::LoadError
false
rescue
Gem.available?(name)
end
Note that the new method allows you to pass a specific version to see if that's loaded:
Gem::Specification.find_by_name('rails', '3.0.4')
You could:
begin
require "somegem"
rescue LoadError
# not installed
end
This wouldn't, however, tell you if the module was installed through gem or some other means.
I use this code and it works smoothly.
def gem_available?(gem_name, version = nil)
version.nil? gem(gem_name) : gem(gem_name, version)
rescue Gem::LoadError
false
end
Examples to use
Let's assume you have rack 1.9.1 installed.
puts gem_available?('rack') # => true
puts gem_available?('rack', '>=2') => # false
Didn't see this anywhere here, but you can also pass fuzzy version strings to find_by_name and find_all_by_name:
Gem::Specification.find_all_by_name('gemname', '>= 4.0').any?

Resources