Bundler: why does it read the gemspec on require "bundler/setup"? - ruby

The title is the question, and here's the context that prompts it.
The Gemfile:
source "http://rubygems.org"
# Specify your gem's dependencies in the gemspec
gemspec
Here is the top of the rackup file:
require 'rubygems'
require "bundler/setup"
On running the rackup file an error is thrown:
<module:Rack>': GemName is not a class (TypeError)
Why? Because I'm writing a piece of Rack middleware, and the standard layout is:
lib/
rack/
gem_name.rb
gem_name/
version.rb
gem_name.rb will contain:
module Rack
class GemName
version.rb will contain:
module Rack
module GemName
VERSION = "0.0.1"
Finally, the gem_name.gemspec will contain:
require "rack/flash-in-the-pan/version"
#...
s.version = Rack::GemName::VERSION
Naming a module and a class by the same name isn't a problem as long as you don't require both files at the same time. Normally, this wouldn't happen, as you either need the version for building the gem, or you need to run the gem library, only one or other gets required.
But, this time I decided to use Bundler to manage the gem's dependencies. When requiring the gem library via bundler it obviously runs the gemspec too. I can "fix" it easily enough, I define the version number by hand in the gemspec.
So back to my question - why does Bundler need to look in the gemspec at the library's runtime?
bundler (1.0.21)
Any insight is much appreciated.

Whenever you run bundler, it has to parse the Gemfile to actually figure out what gems need to be loaded, what has to be added to $LOAD_PATH and so on. As part of that, it has to parse gemspec.
The Gemfile.lock contains info on all of the gems as well as the dependencies to save startup time, but it doesn't alleviate the need for it to parse the Gemfile.
There are various ways you could work around it. Two simple ones would be to use File.read and some regex to pull out the version. Or require the gem_name.rb and gem_name/version.rb files.

Related

When is the 'require' necessary when using a ruby gem?

I noticed for some gems you must include it in the file where you want to use it like this require 'a_gem', but this is not always the case.
I am going to compose a gem by myself. What should I do if I do not want to add the require 'my_gem' to the .rb file when using it?
Usually, an application that is using a gem needs to require the gem:
require "my_awesome_gem"
MyAwesomeGem.do_something_great
However, if an application is using bundler, which defines the application's gem in a file called "Gemfile":
source 'http://rubygems.org'
gem 'my_awesome_gem'
then the application may invoke bundler in a way that automatically requires all of the gems specified in the Gemfile:
require "bundler"
Bundler.require
MyAwesomeGem.do_something_great
Rails projects use Bundler.require, so a Rails application will not need to explicitly require a gem in order to use it: Just add the gem to the Gemfile and go.
For more about Bundler.require, see the bundler documentation
For more about how Rails uses Bundler, see How Does Rails Handle Gems? by Justin Weiss.
This doesn't make sense. If you want to write a Gem and use it, it needs to be required somewhere.
This "somewhere" could be explicit in one of your scripts, it could be written in a Gemfile or it could be required by another script/gem that is required by your script.
If you write a gem, Ruby will not include it automatically in all your scripts.
Finally, if you publish it, should every single Ruby script on earth magically require it and execute your code?
I suppose that the project you have seen which didn't use require 'a_gem' was using a Gemfile.

Why can't I require a gem from within the .gemspec?

I can't require third party code in my .gemspec file. I would like to require a semver library to manage my version number. Why can't I require gems inside my .gemspec?
I have a workaround for my particular problem. What I am interested in the specific reason I can't require code from inside a gemspec. I feel like I am missing something about how Rubygems work. Thanks!
lib = File.expand_path('../lib', __FILE__)
$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
require 'semver'
Gem::Specification.new do |spec|
spec.add_dependency "semver2", "~> 3.4.2"
spec.version = SemVer.find.to_s[1..-1]
end
Running bundle install results in the following error:
$ bundle
[!] There was an error parsing `Gemfile`: There was a LoadError while loading example.gemspec:
cannot load such file -- semver from
/opt/pose/auto/stack-overflow/example/example.gemspec:3:in `<main>'
Obviously there is some point before which you cannot depend on 3rd party gems to be on your LOAD_PATH. I am asking for an explanation of where that point is. My guess is because my Gemfile references my gemspec and Bundler had not loaded all the dependencies yet, but that is a guess.
A .gemspec is just a Ruby script like any other Ruby script. Basically the only thing that is "special" is that it is run in an environment where certain libraries have already been required, such as rubygems/specification.
requireing libraries works just like in any other Ruby script.

Handling require in a project meant to be provided as a gem

I'm completely lost with ruby "require" (and I don't have the option for require_relative since my code is meant to run with ruby 1.8.7.
First some sample code.
requester.rb
require './column/main_column'
....
column/main_column.rb
require './helper' # helper.rb being in the root dir
....
column/segment_column
require './column/main_column'
require './helper' # helper.rb being in the root dir
....
I'm completely lost in how should I use my requires to make my code usable inside a gem and in a standalone way (without assuming a previous gem installation). I was reading at Jekyll's code and every dependencies seems to be loaded in the main rb file which to my opinion is a bit dirty and still do resolve the problem if column/segment_column.rb needs to be called independently.
Please try to argument about different approaches and why one is the best.
You should require files without the extension and relative to your gem project's lib folder.
For example, given this structure:
example-gem/
lib/
example/
gem.rb
example.rb
example-gem.gemspec
You'd require the main example module using:
require 'example'
And in the example.rb file you'd find:
require 'example/gem'
This set of conventions is adopted by the community and is documented on the RubyGems Guides. If they fail, it is often because your gem's lib directory is not in Ruby's $LOAD_PATH, which usually means the gem is not installed.
Building and installing gems during development is cumbersome. For that reason, gems like Bundler became part of the toolboxes of many Rubyists. Bundler lets you use your gem without having to build or install it locally. You can simply bundle exec a script that uses your gem or use bundle console in order to interact with it.

Created a simple gem, but irb can't see it

I created a simple gem by doing:
testgem.gemspec
Gem::Specification.new do |s|
s.name = 'testgem'
s.version = '0.0.1'
s.summary = "code generation"
s.authors = ['asdf']
end
created the following file layout:
testgem.gemspec
Gemfile
lib/
lib/testgem.rb
lib/testgem/other.rb
ran the following:
gem build testgem.gemspec
WARNING: no description specified
WARNING: no email specified
WARNING: no homepage specified
Successfully built RubyGem
Name: testgem
Version: 0.0.1
File: testgem-0.0.1.gem
installed it:
gem install ./testgem-0.0.1.gem
Successfully installed testgem-0.0.1
1 gem installed
tested to see if irb can see it:
irb
require 'testgem'
oadError: cannot load such file -- testgem
I'm using Ruby 1.9.3-p194
It seems to be installed if I do:
gem list
..
testgem (0.0.1)
..
What could be the issue?
You need to add your lib to the search path. Just add
$:.push File.expand_path("../lib", __FILE__)
on top of your .gemspec
When you use require in Ruby, you are trying to load a file, not a gem. The file could of course be contained in a gem, and by convention a gem will have a file with the same name as the gem itself so that if you wanted to use the gem foo you would use require 'foo'. This isn’t always the case though, for example to use the bcrypt-ruby gem you need to require 'bcrypt'.
Additionally a gem doesn’t necessarily have to contain any filles at all. An example of this is the rails gem, which doesn’t contain any files itself (at least in version 3.2), but has dependencies to the various Rails components, so that you can install them all in one step.
In your case, although you have a lib/testgem.rb file in your gems project directory, you are not including it in your gem. You need to specify which files should be included, the default is to not include anything.
To simply include all the files in your project directory, you can add something like:
s.files = Dir.glob '**/*'
to your gemspec.
You need to require 'rubygems' first.

Writing a Jeweler Rakefile that adds dependencies depending on RUBY_ENGINE (ruby or jruby)

I have a Rakefile that includes this:
Jeweler::Tasks.new do |gem|
# ...
gem.add_dependency('json')
end
The gemspec that this generates builds a gem that can't be installed on jruby because the 'json' gem is native.
For jruby, this would have to be:
Jeweler::Tasks.new do |gem|
# ...
gem.add_dependency('json-jruby')
end
How do I conditionally add the dependency for 'json-jruby' when RUBY_ENGINE == 'java'?
It seems like my only option is to manually edit the gemspec file that jeweler generates to add the RUBY_ENGINE check. But I'd rather avoid this, since it kind of defeats the purpose of using jeweler in the first place.
Any ideas?
After some more digging around, I've come to the conclusion that the problem here isn't so much the limitation in Jeweler as it is a problem with how the json-jruby gem is named.
In a perfect world, the json-jruby gem should be named just json. That way when the gem file is built under JRuby (i.e. jgem build json-jruby.gem), the resulting gem file would be named json-1.4.3-universal-java-1.6.gem (RubyGems names the file based on the 'platform' property in the gemspec). When installed with the appropriate platform in the gem name, everything works transparently. There's no need to explicitly require json-jruby. RubyGems uses the appropriate version automatically.
I've contacted the maintainer of the json gem to see if he'd be willing to get json-1.4.3-universal-java-1.6.gem added to his gemcutter repo, but in the meantime you can download the drop-in jruby json gem here: http://roughest.net/json-1.4.3-universal-java-1.6.gem

Resources