Background: Not the Question
I have a project that is running in both development and production in a chruby environment, where ruby-install was installed as root and rubies are stored in /opt/rubies. I have a really hard time (as many people do) getting nokogiri to compile its native extensions as part of a bundle, but it compiles fine as any chruby user outside of the project directory and so long as there are no binstubs. That's just background context; it's not really what my question is about.
My Current Hack
I find that when I have committed binstubs to my project, all kinds of badness happens. So far, my best solution seems to be:
# skip any bundled/binstub version of bundler
rm -rf ./bin
`which bundle` install --binstubs --without="development test"
which seems to work just fine. Nokogiri is apparently not a production dependency (yay!) and the ./bin directory gets trimmed down to just the gems needed for production.
Question: Would the Bundler Deployment Flag Fix Anything?
So, here's the question: What's the difference between what I'm doing here, and bundle install --deploy with or without binstubs? I know it points to vendor/bundle instead of bin by default, but the docs don't really explain the pros and cons of this approach (at least not in a way that I'm understanding).
Aside from knowledge, which is valuable for its own sake, I'd really like to vendor in production gems in a way that it's likely to work across systems (e.g. RHEL6 and RHEL7) without having to rebuild native extensions or strip out development/test gems on production machines.
Is the deployment flag the answer? Or is there a better way to vendor gems with native extensions for cross-distro projects?
I come from a Python and JavaScript background.
When developing a JavaScript project, dependencies are installed in a node_modules directory in the project root.
When developing Python project, typically virtualenvwrapper is used. In this case dependencies are installed in a virtual environment, which is located in ~/.virtualenvs/<project_name> by default.
Now I need to use a ruby tool for a project. The tool that appears to be the most promising for a similar setup as described above, is bundler.
However, the default installation location for bundler is system-wide. I consider this to be harmful.
For one of my systems, it will prompt for a password, at which point I can still abort.
However, for my other system I can write into the global ruby installation. I'm using a homebrew installed ruby here. Bundle will just install dependencies globally.
I know I can specify the installation location by adding --path, but this is easy to forget.
One way to enforce an installation path is by committing .bundle/config. It would just have to contain this:
---
BUNDLE_PATH: "."
However, some googling around shows that it's not adviced to commit this file.
What is the recommended way to prevent accidental global installations using bundler?
Who's to say it will be accidental? It really depends on what context you're talking about here. I have my Ruby set up so that bundle install works without requiring sudo, it's all done through rbenv automatically. The same is true with rvm if done as a user-level install.
When it comes to deploying apps and you want to make sure it's deployed correctly, that's where tools like Capistrano come into play: Create a deployment script that will apply the correct procedure every time.
Checking in a .bundle/config is really rude from a dev perspective, just like checking in any other user-specific preferences you might have. It causes no end of conflict with other team members.
Every time I run any gem command on the command line, Bundler insists on touching my Gemfile.lock file to add this line:
RUBY VERSION
ruby 2.2.2p95
I don't want to commit this to our repository, because it means every dev using a different patch level of Ruby 2.2.2 is going to be in a commit war with me. (I've already resigned myself to a similar issue with the BUNDLED_WITH line.) But I can't deploy unless I do commit that line, because our deploy runs via a rake task and running the deploy leads Bundler to add this block, whereupon the deploy process says, "WAIT! Your working tree is dirty! You might be deploying incomplete changes!!!!1!" (Well, not literally, but you get the idea.)
Can I tell Bundler to leave the RUBY VERSION (and, ideally, BUNDLED_WITH) out of the Gemfile.lock so we don't have to do this ridiculous dance?
(how to prevent bundler from adding platform info to Gemfile.lock seems to be the same question, but there's no answer, natch.)
I don't think so, but maybe it's okay:
As of 2.1.0, Ruby no longer has multiple patch level releases for a given version. See accepted answer on How do version numbers work for MRI Ruby?
2.2.2p95 is the only patch level of 2.2.2 that will ever be released. 'p95' just means that there have been 95 commits since 2.2.0.
Since your whole team will be on 2.2.2 anyway, it shouldn't cause problems to leave this in your Gemfile.lock. (As long as everyone updates Bundler to the version that does this, anyway. Otherwise there'll still be conflicts as the ruby version is added and removed.)
No, it can't be removed, at least in the version(s) of Bundler current as I write this.
This block is added in the #to_lock method of Bundler::Definition. The only conditional it's wrapped in is if locked_ruby_version, and locked_ruby_version is a method which returns either the version defined in an existing lockfile (Gemfile.lock) or the system Ruby - Bundler tries very hard to avoid letting locked_ruby_version return a falsy value.
If you are using some sort of version control like Git you can add the Gemfile.lock to your .gitignore file.
This way anytime you push up a new change you will not push up your Gemfile.lock. Especially since other developers do not need it - once they run bundle, bundle will create a new Gemfile.lock on completion.
I have some gems that are only used for the asset pipeline. One example is:
gem 'jquery-datatables-rails', github: 'rweng/jquery-datatables-rails'
Unfortunately, I can not find exactly where this gem is installed. "gem list --local" does not even show it.
I need to fix it, because I am trying to use Bootstrap styling in datatables, which is allowed in the latest version. But the version of datatables included with the gem is old.
Does anybody know where these gems go? I am very, very confused by the asset pipeline.
I such cases, I fork the project on github and make my changes, and adjust my Gemfiles accordingly. This also makes it reuseable in different projects.
The asset pipeline and Bundler grouping has nothing to do with where gems are installed on your system. You can always run bundle open gemname to open the source of a Gem in your $EDITOR and make quick changes (i.e. for debugging). If you want to actually include changes in a release, though, you are going to want to fork the Gem and make your changes there, then specify the git path in your Gemfile.
As a side note, make sure you run bundle install (or really, just bundle) after making changes to your Gemfile to ensure the Gems all get installed.
I'm sort of new to bundler and the files it generates. I have a copy of a git repo from GitHub that is being contributed to by many people so I was surprised to find that bundler created a file that didn't exist in the repo and wasn't in the .gitignore list.
Since I have forked it, I know adding it to the repo won't break anything for the main repo, but if I do a pull request, will it cause a problem?
Should Gemfile.lock be included in the repository?
Update for 2022 from TrinitronX
Fast-forward to 2021 and now Bundler docs [web archive] now say to commit the Gemfile.lock inside a gem... ¯_(ツ)_/¯ I guess it makes sense for developers and ease of use when starting on a project. However, now CI jobs need to be sure to remove any stray Gemfile.lock files to test against other versions.
Legacy answer ~2010
Assuming you're not writing a rubygem, Gemfile.lock should be in your repository. It's used as a snapshot of all your required gems and their dependencies. This way bundler doesn't have to recalculate all the gem dependencies each time you deploy, etc.
From cowboycoded's comment below:
If you are working on a gem, then DO NOT check in your Gemfile.lock. If you are working on a Rails app, then DO check in your Gemfile.lock.
Here's a nice article explaining what the lock file is.
The real problem happens when you are working on an open-source Rails app that needs to have a configurable database adapter. I'm developing the Rails 3 branch of Fat Free CRM.
My preference is postgres, but we want the default database to be mysql2.
In this case, Gemfile.lock still needs be checked in with the default set of gems, but I need to ignore changes that I have made to it on my machine. To accomplish this, I run:
git update-index --assume-unchanged Gemfile.lock
and to reverse:
git update-index --no-assume-unchanged Gemfile.lock
It is also useful to include something like the following code in your Gemfile. This loads the appropriate database adapter gem, based on your database.yml.
# Loads the database adapter gem based on config/database.yml (Default: mysql2)
# -----------------------------------------------------------------------------
db_gems = {"mysql2" => ["mysql2", ">= 0.2.6"],
"postgresql" => ["pg", ">= 0.9.0"],
"sqlite3" => ["sqlite3"]}
adapter = if File.exists?(db_config = File.join(File.dirname(__FILE__),"config","database.yml"))
db = YAML.load_file(db_config)
# Fetch the first configured adapter from config/database.yml
(db["production"] || db["development"] || db["test"])["adapter"]
else
"mysql2"
end
gem *db_gems[adapter]
# -----------------------------------------------------------------------------
I can't say if this is an established best practice or not, but it works well for me.
My workmates and I have different Gemfile.lock, because we use different platforms, windows and mac, and our server is linux.
We decide to remove Gemfile.lock in repo and create Gemfile.lock.server in git repo, just like database.yml. Then before deploy it on server, we copy Gemfile.lock.server to Gemfile.lock on server using cap deploy hook
Agreeing with r-dub, keep it in source control, but to me, the real benefit is this:
collaboration in identical environments (disregarding the windohs and linux/mac stuff). Before Gemfile.lock, the next dude to install the project might see all kinds of confusing errors, blaming himself, but he was just that lucky guy getting the next version of super gem, breaking existing dependencies.
Worse, this happened on the servers, getting untested version unless being disciplined and install exact version. Gemfile.lock makes this explicit, and it will explicitly tell you that your versions are different.
Note: remember to group stuff, as :development and :test
Simple answer in the year 2021:
Gemfile.lock should be in the version control also for Rubygems. The accepted answer is now 11 years old.
Some reasoning here (cherry-picked from comments):
#josevalim https://github.com/heartcombo/devise/pull/3147#issuecomment-52193788
The Gemfile.lock should stay in the repository because contributors and developers should be able to fork the project and run it using versions that are guaranteed to work.
#rafaelfranca https://github.com/rails/rails/pull/18951#issuecomment-74888396
I don't think it is a good idea to ignore the lock file even for plugins.
This mean that a "git clone; bundle; rake test" sequence is not guarantee to be passing because one of yours dozens of dependencies were upgraded and made your code break. Also, as #chancancode said, it make a lot harder to bisect.
Also Rails has Gemfile.lock in git:
https://github.com/rails/rails/commit/0ad6d27643057f2eccfe8351409a75a6d1bbb9d0
The Bundler docs address this question as well:
ORIGINAL: http://gembundler.com/v1.3/rationale.html
EDIT: http://web.archive.org/web/20160309170442/http://bundler.io/v1.3/rationale.html
See the section called "Checking Your Code into Version Control":
After developing your application for a while, check in the
application together with the Gemfile and Gemfile.lock snapshot. Now,
your repository has a record of the exact versions of all of the gems
that you used the last time you know for sure that the application
worked. Keep in mind that while your Gemfile lists only three gems
(with varying degrees of version strictness), your application depends
on dozens of gems, once you take into consideration all of the
implicit requirements of the gems you depend on.
This is important: the Gemfile.lock makes your application a single
package of both your own code and the third-party code it ran the last
time you know for sure that everything worked. Specifying exact
versions of the third-party code you depend on in your Gemfile would
not provide the same guarantee, because gems usually declare a range
of versions for their dependencies.
The next time you run bundle install on the same machine, bundler will
see that it already has all of the dependencies you need, and skip the
installation process.
Do not check in the .bundle directory, or any of the files inside it.
Those files are specific to each particular machine, and are used to
persist installation options between runs of the bundle install
command.
If you have run bundle pack, the gems (although not the git gems)
required by your bundle will be downloaded into vendor/cache. Bundler
can run without connecting to the internet (or the RubyGems server) if
all the gems you need are present in that folder and checked in to
your source control. This is an optional step, and not recommended,
due to the increase in size of your source control repository.
No Gemfile.lock means:
new contributors cannot run tests because weird things fail, so they won't contribute or get failing PRs ... bad first experience.
you cannot go back to a x year old project and fix a bug without having to update/rewrite the project if you lost your local Gemfile.lock
-> Always check in Gemfile.lock, make travis delete it if you want to be extra thorough https://grosser.it/2015/08/14/check-in-your-gemfile-lock/
A little late to the party, but answers still took me time and foreign reads to understand this problem. So I want to summarize what I have find out about the Gemfile.lock.
When you are building a Rails App, you are using certain versions of gems in your local machine. If you want to avoid errors in the production mode and other branches, you have to use that one Gemfile.lock file everywhere and tell bundler to bundle for rebuilding gems every time it changes.
If Gemfile.lock has changed on your production machine and Git doesn't let you git pull, you should write git reset --hard to avoid that file change and write git pull again.
The other answers here are correct: Yes, your Ruby app (not your Ruby gem) should include Gemfile.lock in the repo. To expand on why it should do this, read on:
I was under the mistaken notion that each env (development, test, staging, prod...) each did a bundle install to build their own Gemfile.lock. My assumption was based on the fact that Gemfile.lock does not contain any grouping data, such as :test, :prod, etc. This assumption was wrong, as I found out in a painful local problem.
Upon closer investigation, I was confused why my Jenkins build showed fetching a particular gem (ffaker, FWIW) successfully, but when the app loaded and required ffaker, it said file not found. WTF?
A little more investigation and experimenting showed what the two files do:
First it uses Gemfile.lock to go fetch all the gems, even those that won't be used in this particular env. Then it uses Gemfile to choose which of those fetched gems to actually use in this env.
So, even though it fetched the gem in the first step based on Gemfile.lock, it did NOT include in my :test environment, based on the groups in Gemfile.
The fix (in my case) was to move gem 'ffaker' from the :development group to the main group, so all env's could use it. (Or, add it only to :development, :test, as appropriate)