Hi,

it seems that thin doesn't like the way that rubygems-bundler handles ARGV. I've put a puts ARGV.inspect at the first start of bin/bundler_wrapper:

{{{
% thin -R faye.ru -C config/faye/development.yml start
["/home/nono/.rvm/gems/ruby-head-patched/bin/thin", "-R", "faye.ru", "-C", "config/faye/development.yml", "start"]
Starting server on tmp/sockets/development.faye.0.sock ...
["start", "--chdir=/home/nono/af83/project", "--environment=development", "--timeout=60", "--log=log/faye.0.log", "--pid=tmp/pids/faye.0.pid", "--max-conns=1024", "--max-persistent-conns=512", "--require=rubygems", "--rackup=faye.ru", "--socket=tmp/sockets/development.faye.0.sock", "--daemonize"]
/home/nono/.rvm/gems/ruby-head-patched/bin/bundler_wrapper:23:in read': No such file or directory - start (Errno::ENOENT) from /home/nono/.rvm/gems/ruby-head-patched/bin/bundler_wrapper:23:in

1. we can handle more then just rubygems-bundler
2. rubygems-bundler can be uninstalled without breaking binaries
3. proof of concept for plugin system for executables for rubygems

My environment:

• OS X Lion
• rvm-managed ruby 1.9.2-p290

From a clean install w/o gems in the current rvm gemset:

$ gem install rubygems-bundler
$ gem install bundler
$ gem install rake

$ head -1 `which bundle`
#!/usr/bin/env bundler_wrapper
$ head -1 `which rake`
#!/usr/bin/env bundler_wrapper

$ echo -e "source 'http://rubygems.org'\ngem 'compass'" > Gemfile

$ bundle

$ head -1 `which sass`
#!/usr/bin/env ruby
$ head -1 `which compass`
#!/usr/bin/env ruby

The 'ruby_noexec_wrapper' needs to honor the --program-suffix passed into ./configure when building ruby.

You can get it from Gem. Here's an example for ruby compiled with --program-suffix=20

Gem.default_exec_format # => "%s20"

Otherwise, if this gem is installed for both (using the above example) ruby and ruby20 then all gem executables will try to run the ruby_noexec_wrapper for whichever ruby is first in the path.

Ciao!

rubygems-bundler/noexec.rb:60:in rubygems_spec': undefined methodexecutables' for #Array:0x7f291216dcc0 (NoMethodError)

Instead of requiring the user run gem regenerate_binstubs, you could automatically invoke it using gemspec.extensions to run a post-install script.

@mpapis

I can't use gems when I'm in a project, that exist in global. they positively HAVE to be in the Gemfile, or I can't use them. Is that supposed to happen? (Blocking global gems if they're NOT in the Gemfile?)

I thought it was supposed just cause limitation of use to the version of gems listed in Gemfile.lock, not block those not listed all together. https://gist.github.com/2338630 - I have github gem in 1.9.3-p125@global gemset. Its NOT listed in the project's Gemfile. I should be able to run that gem from the commandline without rubygems-bundler blocking its use since its NOT in the Gemfile.

e.g but it SHOULD be selecting the version of tzinfo i DO have in the Gemfile.lock (i have 2 versions of tzinfo installed,
the older one is in the Gemfile.lock) and that it does do correctly.

So basically, in cucumber feature format:

Given a gem is not in a project's Gemfile
And the gem is in the global gemset
Then rubygems-bundler should not attempt to execute it within the project's context
And rubygems-bundler should not block its command-line use

I can understand what "ruby_bundler_wrapper" means, but I have no idea what "ruby_noexec_wrapper" means. After searching online, I'm guessing it's a reference to https://github.com/joshbuddy/noexec.

I think "ruby_bundler_wrapper" is clearer, or even "ruby_bundle_exec_wrapper". But if you like "ruby_noexec_wrapper", please explain what noexec is.

Currently, if the version specified in the Gemfile for the target gem is not installed, then noexec gives up and allows Ruby to select the default version. IMHO, this defeats the whole purpose of using bundle exec. It should fail fast (or make a setting to determine whether it does or not).

To reproduce, install a gem. Then, create a Gemfile that specifies it as a requirement.

gem "sass", "3.1.21"

Then, get it installed.

bundle install

Next, change the version to something different.

gem "sass", "3.1.20"

Finally, run the binary and consult the version.

sass --version

You'll see that it reports "3.1.21" instead of failing to execute.

Turning on debug, you'll see that noexec wasn't happy with what was going on:

$ NOEXEC_DEBUG=1 sass --version
Noexec - starting check
Examining /home/dallen/tmp/foo/Gemfile
Considering "/home/dallen/tmp/foo/.noexec.yaml"
Ignoring candidate /home/dallen/tmp/foo/Gemfile:
Could not find gem 'sass (= 3.1.20) ruby' in the gems available on this machine.
No valid Gemfile found, moving on
Sass 3.1.21 (Brainy Betty)

I think this should be a failure scenario.

$ ruby -v
ruby 1.9.2p180 (2011-02-18 revision 30909) [i686-linux]
$ gem -v
1.8.5
$ gem update
Updating installed gems
Updating rubygems-bundler
Building native extensions. This could take a while...
ERROR: Error installing rubygems-bundler:
ERROR: Failed to build gem native extension.

    /usr/bin/ruby1.9.1 extconf.rb

extconf.rb:15:in join': can't convert nil into String (TypeError) from extconf.rb:15:in

Not sure why, but when i regenerated by binstubs, it skipped spork for me. Here is my gemfile:

https://gist.github.com/2421841

But it clearly does not generate a binstub for spork. Here's the output:

archive-tar-minitar 0.5.2
bundler 1.1.3
coderay 1.0.5
cucumber 1.1.9
cucumber 1.1.4
debugger 1.0.1
diff-lcs 1.1.3
erubis 2.7.0
fog 1.3.1
fog 1.1.2
git_remote_branch 0.3.3
haml 3.1.4
heroku 2.24.1
heroku 2.14.0
hirefireapp 0.0.5
hoe 1.5.3
httparty 0.8.1
km 1.1.1
launchy 2.1.0
launchy 2.0.5
newrelic_rpm 3.3.0
nokogiri 1.5.2
nokogiri 1.5.0
oauth 0.4.5
oink 0.9.3
parallel_tests 0.6.17
pry 0.9.8.2
rack 1.4.1
railties 3.2.3
railties 3.2.2
rake 0.9.2
rdiscount 1.6.8
rdoc 3.12
resque 1.20.0
rest-client 1.6.7
rspec-core 2.7.1
ruby-graphviz 0.9.21
ruby-prof 0.10.8
rubyforge 2.0.4
rubygems-bundler 0.9.0
sass 3.1.15
sass 3.1.12
tddium 1.0.1
term-ansicolor 1.0.7
thin 1.3.1
thor 0.14.6
tilt 1.3.3
treetop 1.4.10
vanity 1.8.0
xray 1.1

This is related to issue #15. When using gem install --env-shebang, the process fails with:

ERROR:  While executing gem ... (NoMethodError)
    undefined method `opts' for #

The fix for this issue is:

--- /tmp/rubygems_bundler_installer.rb.old  2012-05-05 19:01:27.000000000 +0200
+++ /tmp/rubygems_bundler_installer.rb  2012-05-05 19:01:37.000000000 +0200
@@ -54,7 +54,7 @@
 
     if not ruby_name then
       "#!#{Gem.ruby}#{opts}"
-    elsif inst.opts then
+    elsif opts then
       "#!/bin/sh\n'exec' #{ruby_name.dump} '-x' \"$0\" \"$@\"\n#{shebang}"
     else
       @env_path ||= Gem::Installer::ENV_PATHS.find {|env_path| File.executable? env_path } 

I'm seeing an undefined method `name' for # when I attempt to deploy that I haven't seen before. I have pasted the output of my deploy below in the gist.

rvm ruby-1.8.7-p334 do gem --version => 1.3.6

https://gist.github.com/mtmcfarl/8c21a05447b9124ac7b4

as per @mpapis

Hi there,

I'm wondering if I'll have to run 'gem regenerate_binstubs' every time a new gem is installed...it seems so, although I don't like the idea very much :)

Here's a recipe if you want to see what I mean. On bash, run:

rvm --create 1.9.2@rubygems-bundler_test
mkdir ~/rubygems-bundler_test
cd ~/rubygems-bundler_test
gem instal rubygems-bundler
echo -e "custom_shebang: $env ruby_bundler_wrapper" > ~/.gemrc
gem install bundler
gem regenerate_binstubs
gem install haml
echo -e "source 'http://rubygems.org'\ngem 'haml', '3.0.0'" >> Gemfile
bundle
USE_BUNDLER=no haml -v
USE_BUNDLER=try haml -v
bundle exec haml -v
gem regenerate_binstubs
USE_BUNDLER=no haml -v
USE_BUNDLER=try haml -v
bundle exec haml -v

You'll see that the first time I run 'USE_BUNDLER=try haml -v' it does not work, but after a new 'gem regenerate_binstubs' it works really well.

Did I get it right? If so, are you guys planning to address this issue?

Happy new year! :D

Best,
Daniel

Hi,

On my fresh OSX Maverick installation, I can't execute rubygems-bundler-uninstaller:

$ sudo rubygems-bundler-uninstaller
Password:
sudo: rubygems-bundler-uninstaller: command not found

Note that I am forced to run bundle exec.

Here, more informations:

$ gem env
RubyGems Environment:
  - RUBYGEMS VERSION: 2.2.1
  - RUBY VERSION: 2.0.0 (2013-06-27 patchlevel 247) [universal.x86_64-darwin13]
  - INSTALLATION DIRECTORY: /Library/Ruby/Gems/2.0.0
  - RUBY EXECUTABLE: /System/Library/Frameworks/Ruby.framework/Versions/2.0/usr/bin/ruby
  - EXECUTABLE DIRECTORY: /usr/bin
  - SPEC CACHE DIRECTORY: /Users/mehdi/.gem/specs
  - RUBYGEMS PLATFORMS:
    - ruby
    - universal-darwin-13
  - GEM PATHS:
     - /Library/Ruby/Gems/2.0.0
     - /Users/mehdi/.gem/ruby/2.0.0
     - /System/Library/Frameworks/Ruby.framework/Versions/2.0/usr/lib/ruby/gems/2.0.0
  - GEM CONFIGURATION:
     - :update_sources => true
     - :verbose => true
     - :backtrace => false
     - :bulk_threshold => 1000
     - "gem" => "--no-ri --no-rdoc"
  - REMOTE SOURCES:
     - https://rubygems.org/
  - SHELL PATH:
     - /usr/local/share/npm/bin
     - /usr/local/bin
     - /Users/mehdi/Applications/Scripts
     - /usr/bin
     - /bin
     - /usr/sbin
     - /sbin
     - /usr/local/bin

$ gem list

*** LOCAL GEMS ***

bundler (1.5.2)
bundler-unload (1.0.2)
CFPropertyList (2.2.0)
executable-hooks (1.3.1)
libxml-ruby (2.6.0)
nokogiri (1.5.6)
rubygems-bundler (1.4.2)
rubygems-update (2.2.1)
sqlite3 (1.3.7)

$ echo rubygems-bundler executable-hooks bundler-unload bundler$ | xargs -n 1 gem list -d

*** LOCAL GEMS ***

rubygems-bundler (1.4.2)
    Authors: Josh Hull, Michal Papis
    Homepage: http://mpapis.github.com/rubygems-bundler
    License: Apache 2.0
    Installed at: /Library/Ruby/Gems/2.0.0

    Stop using bundle exec

*** LOCAL GEMS ***

executable-hooks (1.3.1)
    Author: Michal Papis
    Homepage: https://github.com/mpapis/executable-hooks
    License: Apache 2.0
    Installed at: /Library/Ruby/Gems/2.0.0

    Hook into rubygems executables allowing extra actions to be taken
    before executable is run.

*** LOCAL GEMS ***

bundler-unload (1.0.2)
    Author: Michal Papis
    Homepage: https://github.com/mpapis/bundler-unload
    License: Apache 2.0
    Installed at: /Library/Ruby/Gems/2.0.0

    Allow unloading bundler after Bundler.load

*** LOCAL GEMS ***

bundler (1.5.2)
    Authors: André Arko, Terence Lee, Carl Lerche, Yehuda Katz
    Homepage: http://bundler.io
    License: MIT
    Installed at: /Library/Ruby/Gems/2.0.0

    The best way to manage your application's dependencies

rubygems-bundler (1.4.2)
    Authors: Josh Hull, Michal Papis
    Homepage: http://mpapis.github.com/rubygems-bundler
    License: Apache 2.0
    Installed at: /Library/Ruby/Gems/2.0.0

    Stop using bundle exec

$ ls -l /Library/Ruby/Gems/2.0.0
total 0
drwxr-xr-x  2 root  wheel   68 20 jan 10:25 build_info/
drwxr-xr-x  7 root  wheel  238 20 jan 10:58 cache/
drwxr-xr-x  2 root  wheel   68 20 jan 10:25 doc/
drwxr-xr-x  3 root  wheel  102 20 jan 10:27 extensions/
drwxr-xr-x  7 root  wheel  238 20 jan 10:58 gems/
drwxr-xr-x  7 root  wheel  238 20 jan 10:58 specifications/

$ ls -l /Users/mehdi/.gem/ruby/2.0.0/
total 0
drwxr-xr-x  3 mehdi  staff  102 20 jan 10:25 cache/

$ ls -l /System/Library/Frameworks/Ruby.framework/Versions/2.0/usr/lib/ruby/gems/2.0.0/
total 0
drwxr-xr-x  3 root  wheel  102 25 aoû 05:33 bin/
drwxr-xr-x  2 root  wheel   68 25 aoû 05:32 build_info/
drwxr-xr-x  6 root  wheel  204 25 aoû 05:33 cache/
drwxr-xr-x  6 root  wheel  204 25 aoû 05:33 doc/
drwxr-xr-x  9 root  wheel  306  6 nov 15:24 gems/
drwxr-xr-x  7 root  wheel  238  6 nov 15:24 specifications/

Hi

Maybe I'm not using BUNDLE_GEMFILE properly but when it's set Bundler.setup is never called. So, in the code for noexec.rb

  def setup
      log "Noexec"
      return if %w(bundle rubygems-bundler-uninstaller).include?(File.basename($0))
      return if ENV['BUNDLE_GEMFILE'] # <--- bail out?
      return if %w(0 skip).include?( ENV['NOEXEC'] )
      gemfile = File.join(CURRENT, "Gemfile")
      while true
        if File.exist?(gemfile)
          log "Examining #{gemfile}"
          if Noexec.candidate?(gemfile, File.basename($0))
            log "Using #{gemfile}"
            ENV['BUNDLE_GEMFILE'] = gemfile
            Bundler.setup # <----- never get here
            return
          end
        end
      #and so on

Is there a reason for breaking out that fast? What I have is a custom Gemfile that's set and this is a gem from git (yes, I realize this is an edge case on top of an edge case).

I got it to work by doing this:

    def setup
      log "Noexec"
      return if %w(bundle rubygems-bundler-uninstaller).include?(File.basename($0))
      # removed return if ENV['BUNDLE_GEMFILE'] 
      return if %w(0 skip).include?( ENV['NOEXEC'] )
      gemfile = ENV['BUNDLE_GEMFILE'] || File.join(CURRENT, "Gemfile") # <-- and put it here

But I don't want to submit a pull request as you are MUCH more familiar with Gem and Bundler internals by now. Am I being dense?

See:
mattray/spiceweasel#23

I just installed Ruby 2.0.0-p247, followed by gem install rubygems-bundler. In p247 it seems that my Gemfile is no longer being honored. In addition, I get warnings about "Unresolved specs".

For example when using p247:

$ rbenv shell 2.0.0-p247 
$ guard
WARN: Unresolved specs during Gem::Specification.reset:
      lumberjack (>= 1.0.2)
WARN: Clearing out unresolved specs.
Please report a bug if this causes problems.
15:07:59 - INFO - Guard here! It looks like your project has a Gemfile, yet you are running
> [#] `guard` outside of Bundler. If this is your intent, feel free to ignore this
> [#] message. Otherwise, consider using `bundle exec guard` to ensure your
> [#] dependencies are loaded correctly.
> [#] (You can run `guard` with --no-bundler-warning to get rid of this message.)
15:07:59 - INFO - Guard uses TerminalNotifier to send notifications.
15:07:59 - INFO - Guard uses TerminalTitle to send notifications.
15:07:59 - INFO - Guard::RSpec is running

But Ruby 2.0.0-p195 is fine:

$ rbenv shell 2.0.0-p195 
$ guard
15:07:50 - INFO - Guard uses TerminalNotifier to send notifications.
15:07:50 - INFO - Guard uses TerminalTitle to send notifications.
15:07:50 - INFO - Guard::RSpec is running

I've run bundle install in both environments, and bundle exec guard works fine with no warnings.

Am I doing something wrong, or is this a problem with rubygems-bundler + 2.0.0-p247?

For the record, here are my ruby and gem versions:

$ rbenv shell 2.0.0-p195 
$ ruby -v
ruby 2.0.0p195 (2013-05-14 revision 40734) [x86_64-darwin12.3.0]
$ gem -v
2.0.2

$ rbenv shell 2.0.0-p247 
$ ruby -v
ruby 2.0.0p247 (2013-06-27 revision 41674) [x86_64-darwin12.4.0]
$ gem -v
2.0.3

Background: https://gist.github.com/swalberg/13cddd3a389b0a1bfaea

I have a Rails repo where I do my development on a Vagrant environment (guest) from an OSX box (host). My Gemfile contains capistrano but not vagrant. After upgrading rvm and ruby on my app, I was unable to run vagrant and cap from within my project directory. I was able to do it from my home directory and another Rails app.

I did an rvm implode and started again. Also removed everything from vendor/ruby in my project.

Running bundle install from the host let me access cap but not vagrant.

Adding exclude: [vagrant] to ./.noexec.yaml fixed the problem temporarily. Also, removing this file and then:

$ rubygems-bundler-uninstaller
archive-tar-minitar 0.5.2
bundler 1.3.4
capistrano 2.14.2
erubis 2.7.0
httparty 0.10.2
rake 10.0.4
#rake 0.9.2.2 not found in GEM_PATH
#rdoc 3.9.5 not found in GEM_PATH
rest-client 1.6.7
rubygems-bundler 1.1.1
vagrant 1.0.7

(seems to have) fixed it permanently.

FWIW I have a coworker who ran into the same problem but a lack of tenacity (or an affinity for the current hawtness) made him move over to rbenv.

I'm trying to run bluepill from inside a directory with deployed gems, but bluepill can't be found from the default ruby environment.

This setup can be duplicated by following the rvm and code-related instructions at https://github.com/discourse/discourse/blob/master/docs/INSTALL-ubuntu.md

Environment details are here:
https://gist.github.com/Supermathie/747d797ba062f33c7b92

I'm debugging an issue with the spring gem failing to load. This has lead me (down a long long trail) to rubygems-bundler. Hopefully the following trace will show the issue

[81, 90] in /Users/andy/.rvm/gems/ruby-1.9.3-p448@global/gems/rubygems-bundler-1.4.1/lib/rubygems-bundler/noexec.rb
   81        if File.file?(gemfile) && candidate?
   82          log "Keeping #{gemfile} loaded"
   83          Bundler.setup
   84          return
   85        end
=> 86        new_gemfile = File.expand_path("../../Gemfile", gemfile)
   87        break if new_gemfile == gemfile
   88        @gemfile = new_gemfile
   89      end
   90      log "No valid Gemfile found, moving on"
(rdb:1) gemfile
"/Users/andy/Sites/futuregov/patchwork/Gemfile"    # <- actual location of Gemfile
(rdb:1) n
/Users/andy/.rvm/gems/ruby-1.9.3-p448@global/gems/rubygems-bundler-1.4.1/lib/rubygems-bundler/noexec.rb:87
break if new_gemfile == gemfile

[82, 91] in /Users/andy/.rvm/gems/ruby-1.9.3-p448@global/gems/rubygems-bundler-1.4.1/lib/rubygems-bundler/noexec.rb
   82          log "Keeping #{gemfile} loaded"
   83          Bundler.setup
   84          return
   85        end
   86        new_gemfile = File.expand_path("../../Gemfile", gemfile)
=> 87        break if new_gemfile == gemfile
   88        @gemfile = new_gemfile                   
   89      end
   90      log "No valid Gemfile found, moving on"
   91    rescue LoadError
(rdb:1) new_gemfile
"/Users/andy/Sites/futuregov/Gemfile"         # <- NO GEMFILE HERE

Now I have no idea why this code is doing what its doing, or even if its wrong (it doesn't look correct), but the fact that Spring is failing with /Gemfile not found, and that somehow this code is executing when Spring runs (it loads bundler) makes me very suspicious.

Perhaps the code should check if the existing gemfile is actually a file before changing its path. Doing this after changing its path might also be an idea.

TIA

Andrew

Hi,

Since a few days ago my gems started giving me errors about ruby_noexec_wrapper.
Now I tried uninstalling rubygems-bundler but the uninstaller fails absurdly

$ NOEXEC_DEBUG=1 rubygems-bundler-uninstaller                                                                                                                                                                                                            
Bin used: /Users/timosand/.rvm/gems/ruby-2.0.0-p247/bin/rubygems-bundler-uninstaller
Bin calculated: rubygems-bundler-uninstaller
Noexec - skipped binary: rubygems-bundler-uninstaller
/Users/timosand/.rvm/rubies/ruby-2.0.0-p247/lib/ruby/site_ruby/2.0.0/rubygems.rb:286:in `bin_path': can't find gem rubygems-bundler ([">= 0"]) with executable rubygems-bundler-uninstaller (Gem::GemNotFoundException)
    from /Users/timosand/.rvm/gems/ruby-2.0.0-p247/bin/rubygems-bundler-uninstaller:23:in `<main>'
    from /Users/timosand/.rvm/gems/ruby-2.0.0-p247/bin/ruby_noexec_wrapper:14:in `eval'
    from /Users/timosand/.rvm/gems/ruby-2.0.0-p247/bin/ruby_noexec_wrapper:14:in `<main>'

Please help quickly, @mpapis

https://github.com/mpapis/noexec

When bundler is invoked with the --deployment option (best practice for deployments) it generates a .bundle/config file which explicitly disables the use of shared gems, i.e. an isolated bundle is to be created.

If rubygems-bundler is installed alongside the bundler gem, bundler will attempt to load it before it installs the gems specified in your Gemfile. But because we are by now operating in this "shared gems are disabled" setup, rubygems-bundler will not be found and the bundle will fail.

In this situation it seems you must completely uninstall rubygems-bundler before invoking bundler, or else avoid using the --deployment option.

Did I say the word "bundle" enough?

If there is a directory named Gemfile, the condition is true, but doomed to fail when bundler notices that it's a directory rather than a file.

I think the more correct implementation is to "require 'pathname'" then check "Pathname.new(gemfile).file?".

It would be great to have an option so it would exclude all the gems I have in rvm global gemset, instead of specifying every binary by hand

The engineyard-gem is not meant to be run inside a bundled environment. However rubygems-bundler doesn't seem to know that and breaks the gem when trying to run it. It would make sense for rubygems-bundler to check if a gem is in the current Gemfile before attempting to run with bundle exec.

Invoking 'vagrant' fails on startup with this message.

To reproduce:

$ vagrant -h
Error loading RubyGems plugin "/home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/rubygems-bundler-1.0.2/lib/rubygems_plugin.rb": Could not locate Gemfile (Bundler::GemfileNotFound)
Error loading RubyGems plugin "/home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/rubygems-bundler-1.0.2/lib/rubygems_plugin.rb": Could not locate Gemfile (Bundler::GemfileNotFound)
Usage: vagrant [-v] [-h] command []

-v, --version                    Print the version and exit.
-h, --help                       Print this help.

Available subcommands:
box
destroy
gem
halt
init
package
provision
reload
resume
ssh
ssh-config
status
suspend
up

For help on any individual command run vagrant COMMAND -h

Additional diagnostics:
$ NOEXEC=0 vagrant -h
Error loading RubyGems plugin "/home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/rubygems-bundler-1.0.2/lib/rubygems_plugin.rb": Could not locate Gemfile (Bundler::GemfileNotFound)
Error loading RubyGems plugin "/home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/rubygems-bundler-1.0.2/lib/rubygems_plugin.rb": Could not locate Gemfile (Bundler::GemfileNotFound)
Usage: vagrant [-v] [-h] command []

-v, --version                    Print the version and exit.
-h, --help                       Print this help.

Available subcommands:
box
destroy
gem
halt
init
package
provision
reload
resume
ssh
ssh-config
status
suspend
up

For help on any individual command run vagrant COMMAND -h

ruby -rbundler -e "puts Bundler.settings[:disable_shared_gems]"
/home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/bundler-1.1.4/lib/bundler/shared_helpers.rb:22:in default_gemfile': Could not locate Gemfile (Bundler::GemfileNotFound) from /home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/bundler-1.1.4/lib/bundler.rb:213:indefault_gemfile'
from /home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/bundler-1.1.4/lib/bundler.rb:164:in root' from /home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/bundler-1.1.4/lib/bundler.rb:170:inapp_config_path'
from /home/karsten/.rvm/gems/ruby-1.9.3-p125@global/gems/bundler-1.1.4/lib/bundler.rb:182:in settings' from -e:1:in

If a Gem is using the :path option in a Gemfile then it gets passed over when iterating through the specs

in regenerate_binstubs_command.rb, circa line 37

        cache_gem = File.join(org_gem_path, 'cache', spec.file_name)
        if File.exist? cache_gem
        # stuff
        else
          puts "##{spec.name} #{spec.version} not found in GEM_PATH" # <--- no love
        end

I don't think :path Gems get their spec copied to the cache... If I manually change the wrapper everything is peachy.

Deploying via rvm-capistrano results in bundle config having the path set to
#{deploy_to}/shared/bundle. Whether rubygems-bundler is installed into the default gemset or global, or to shared/bundle, gem regenerate_binstubs won't generate binstubs for all the commands in #{deploy_to}/shared/bundle:

https://gist.github.com/aspiers/5091419

(I don't think this is specific to rake, but it's the original stimulus here.)

With an rvm-installed ruby and the bundler gem installed, here's a minimal reproduction showing what I mean:

mkdir -p /tmp/foo
cd /tmp/foo

echo 'source :rubygems

gem "sass", "= 99.9"' > Gemfile

rake foo

(The version of sass intentionally high to force a missing gem.)

From the $ rake foo command, there's a stacktrace as the output:

/home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/resolver.rb:287:in `resolve': Could not find gem 'sass (= 99.9) ruby' in the gems available on this machine. (Bundler::GemNotFound)
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/resolver.rb:161:in `start'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/resolver.rb:128:in `block in resolve'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/resolver.rb:127:in `catch'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/resolver.rb:127:in `resolve'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/definition.rb:192:in `resolve'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/definition.rb:127:in `specs'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/gems/bundler-1.1.4/lib/bundler/environment.rb:27:in `specs'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194@global/gems/rubygems-bundler-1.0.2/lib/rubygems-bundler/noexec.rb:41:in `candidate?'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194@global/gems/rubygems-bundler-1.0.2/lib/rubygems-bundler/noexec.rb:60:in `setup'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194@global/gems/rubygems-bundler-1.0.2/lib/rubygems-bundler/noexec.rb:75:in `<top (required)>'
    from /home/adam/.rvm/rubies/ruby-1.9.3-p194/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:60:in `require'
    from /home/adam/.rvm/rubies/ruby-1.9.3-p194/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:60:in `rescue in require'
    from /home/adam/.rvm/rubies/ruby-1.9.3-p194/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:35:in `require'
    from /home/adam/.rvm/gems/ruby-1.9.3-p194/bin/ruby_noexec_wrapper:9:in `<main>'

Contrast that with the output from $ bundle exec rake foo:

Could not find gem 'sass (= 99.9) ruby' in the gems available on this machine.
Run `bundle install` to install missing gems.

To a user who is ignorant of Bundler, the stacktrace doesn't tell you how to fix it, despite it being very simple, whereas the bundle exec version makes it very clear. Since rubygems-bundler seems to intend to make the bundle exec redundant, and a missing gem dependency isn't such an exceptional case, I think it would be better to follow the simpler output instead of the stacktrace. (It also has the added bonus that to someone who doesn't yet fully understand how Bundler works, or when rake tasks are used as part of a non-Ruby project where Ruby ecosystem tools aren't assumed knowledge, it clearly tells you how to fix the problem immediately.)

Hi. I'm having a Psych loading problem in my Rails project (loaded twice). When I run a rake task (or even Zeus), I get the following warnings:

/Users/hugo/.rvm/gems/ruby-1.9.3-p385@score/gems/psych-2.0.1/lib/psych.rb:219: warning: already initialized constant VERSION
/Users/hugo/.rvm/gems/ruby-1.9.3-p385@score/gems/psych-2.0.1/lib/psych.rb:222: warning: already initialized constant LIBYAML_VERSION

I'm using RVM 1.22.15, ruby 1.9.3-p385, rubygems 2.1.5, bundler 1.3.5.

I can fix the issue by downgrading rubygems-bundler to version 1.1.0.

I inspected the source code and I see that noexec.rb requires 'yaml'. This of course is most likely loading the psych.rb included in ruby 1.9.3, which is version 1.3.4.

These are just warnings. But later, if I run some ActiveRecord code, it tries to call Psych::Visitors::YAMLTree.create which does not exist in version 1.3.4.

Short story, is there a way to fix this? Could noexec.rb unload yaml when done with parsing yaml stuff?

this may be WAD but I figured I would point it out

using rubygems 1.7.2

the 0.2.6 gemspec has a date format that is unsupported

gems/1.8/specifications/rubygems-bundler-0.2.6.gemspec]: invalid date format in specification: "2011-09-16 00:00:00.000000000Z"

The source on github just has 2011-09-16 so guess whatever publishes it to rubygems modifies the date field, making it incompatible with older versions of rubygems.

for now I've reverted to 0.2.5.. upgrading to a newer rubygems is not currently an option for my project.

Tony

Trying to jist a file and I get:

$ jist out
/Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/bundler-1.1.5/lib/bundler/dsl.rb:12:in `evaluate': There was an error in your Gemfile, and Bundler cannot continue. (Bundler::GemfileError)
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/bundler-1.1.5/lib/bundler/definition.rb:18:in `build'
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/bundler-1.1.5/lib/bundler.rb:135:in `definition'
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/bundler-1.1.5/lib/bundler.rb:123:in `load'
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/rubygems-bundler-1.0.6/lib/rubygems-bundler/noexec.rb:41:in `candidate?'
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/rubygems-bundler-1.0.6/lib/rubygems-bundler/noexec.rb:60:in `setup'
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370@global/gems/rubygems-bundler-1.0.6/lib/rubygems-bundler/noexec.rb:75
        from /Users/mat/.rvm/rubies/ruby-1.8.7-p370/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:60:in `gem_original_require'
        from /Users/mat/.rvm/rubies/ruby-1.8.7-p370/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:60:in `require'
        from /Users/mat/.rvm/gems/ruby-1.8.7-p370/bin/ruby_noexec_wrapper:9

Also, jist is not in that Gemfile, if it would ever help :-)

Some companies will only use gems with a certain license.
The canonical and easy way to check is via the gemspec
via e.g.

spec.license = 'MIT'
# or
spec.licenses = ['MIT', 'GPL-2']

There is even a License Finder to help companies ensure all gems they use
meet their licensing needs. This tool depends on license information being available in the gemspec.
Including a license in your gemspec is a good practice, in any case.

How did I find you?

I'm using a script to collect stats on gems, originally looking for download data, but decided to collect licenses too,
and make issues for missing ones as a public service :)
https://gist.github.com/bf4/5952053#file-license_issue-rb-L13 So far it's going pretty well

The Installation section of the readme says:

    gem install rubygems-bundler

Next run (once):

    gem regenerate_binstubs

And you're done!

However, I find that rails server still needs to be invoked with the incantation bundle exec rails server:

$ rails server
Rails is not currently installed on this system. To get the latest version, simply type:

    $ sudo gem install rails

You can then rerun your "rails" command.
$ bundle exec rails server
=> Booting WEBrick
=> Rails 3.2.3 application starting in development on http://0.0.0.0:3000
=> Call with -d to detach
=> Ctrl-C to shutdown server
[2012-05-31 05:13:51] INFO  WEBrick 1.3.1
[2012-05-31 05:13:51] INFO  ruby 1.9.3 (2012-04-20) [x86_64-darwin10.8.0]
[2012-05-31 05:13:51] INFO  WEBrick::HTTPServer#start: pid=1220 port=3000
^C[2012-05-31 05:15:30] INFO  going to shutdown ...
[2012-05-31 05:15:30] INFO  WEBrick::HTTPServer#start done.
Exiting
$ gem regenerate_binstubs
bundler 1.1.3
rake 0.9.2.2
rubygems-bundler 1.0.2
$ rails server
Rails is not currently installed on this system. To get the latest version, simply type:

    $ sudo gem install rails

You can then rerun your "rails" command.
$ bundle exec rails server
=> Booting WEBrick
=> Rails 3.2.3 application starting in development on http://0.0.0.0:3000
=> Call with -d to detach
=> Ctrl-C to shutdown server
[2012-05-31 05:13:51] INFO  WEBrick 1.3.1
[2012-05-31 05:13:51] INFO  ruby 1.9.3 (2012-04-20) [x86_64-darwin10.8.0]
[2012-05-31 05:13:51] INFO  WEBrick::HTTPServer#start: pid=1220 port=3000
^C[2012-05-31 05:15:30] INFO  going to shutdown ...
[2012-05-31 05:15:30] INFO  WEBrick::HTTPServer#start done.
Exiting

rails new command creates a Gemfile, but not Gemfile.lock.
It runs a bundle command, but on the parent directory's Gemfile, not the target.

rails new app
ls -l app/Gemfile*

gem regenerate_binstubs as ruby restores the expected behavior

i using system wide gem, console utility, not related with any project.
when i cd into project with gemfile, its broken.

example, my gem confuddle and executable 'un':

gem list confuddle
*** LOCAL GEMS ***
confuddle (0.0.2)

which un
/home/kostya/.rvm/gems/ree-1.8.7-2012.02/bin/un

un
---ok

after cd into some project with gemfile

gem list confuddle
*** LOCAL GEMS ***
confuddle (0.0.2)


which un
/home/kostya/.rvm/gems/ree-1.8.7-2012.02/bin/un

un
/home/kostya/.rvm/rubies/ree-1.8.7-2012.02/lib/ruby/site_ruby/1.8/rubygems.rb:812:in `report_activate_error': Could not find RubyGem confuddle (>= 0) (Gem::LoadError)
    from /home/kostya/.rvm/rubies/ree-1.8.7-2012.02/lib/ruby/site_ruby/1.8/rubygems.rb:223:in `activate'
    from /home/kostya/.rvm/rubies/ree-1.8.7-2012.02/lib/ruby/site_ruby/1.8/rubygems.rb:1146:in `gem'
    from /home/kostya/.rvm/gems/ree-1.8.7-2012.02/bin/un:18
    from /home/kostya/.rvm/gems/ree-1.8.7-2012.02/bin/ruby_noexec_wrapper:14

is it bug of rubygems-bundler or something else?

After installing a gem through "gem install mygem --env-shebang", after the wrapper gets overwritten by the fix-wrapper.rb hook, the shebang is overwritten as @env_shebang is not defined in RubyGemsBundlerInstaller. It should take into account the configuration passed in gem.

Was testing rubygems-bundler and ran into this issue.

Steps to reproduce

1. gem install rake gist rubygems-bundler
2. gem regenerate_binstubs
3. rails new foobar
4. Add gem 'rake', '0.9.6' to Gemfile.
5. bundle install --binstubs
6. rails g resource Baz
7. rake routes | gist

Expected Results

Runs rake 0.9.6 and the latest version of the gist gem.

Actual Results

rake aborted!
You have already activated rake 10.0.4, but your Gemfile requires rake 0.9.6. Using bundle exec may solve this.
/home/hal/.gem/ruby/1.9.1/gems/bundler-1.3.5/lib/bundler/runtime.rb:33:in `block in setup'
/home/hal/.gem/ruby/1.9.1/gems/bundler-1.3.5/lib/bundler/runtime.rb:19:in `setup'
/home/hal/.gem/ruby/1.9.1/gems/bundler-1.3.5/lib/bundler.rb:120:in `setup'
/home/hal/.gem/ruby/1.9.1/gems/bundler-1.3.5/lib/bundler/setup.rb:17:in `'
/home/hal/foobar/config/boot.rb:6:in `'
/home/hal/foobar/config/application.rb:1:in `'
/home/hal/foobar/Rakefile:5:in `'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/rake_module.rb:25:in `load'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/rake_module.rb:25:in `load_rakefile'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:589:in `raw_load_rakefile'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:89:in `block in load_rakefile'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:160:in `standard_exception_handling'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:88:in `load_rakefile'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:72:in `block in run'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:160:in `standard_exception_handling'
/home/hal/.gem/ruby/1.9.1/gems/rake-10.0.4/lib/rake/application.rb:70:in `run'

Documentation bug/feature request:
One can remove the gem, but how does one back-out the changes to the binstubs?
The documentation doesn't say how to uninstall

In a rails project with tmuxinator in gemfile, the only way to run tmuxinator is via "bundle exec".

I've done a "gem regenerate_binstubs" but it has no effect on the "tmuxinator" command - I still need to do "bundle exec tmuxinator".

ruby 2.0.0p247, rubygems-bundler 1.3.2

Please let me know if I can provide more information.

It's pretty easy to reproduce. Create a Gemfile with this content:

gem "jammit", "0.6.5", git: "https://github.com/documentcloud/jammit.git"

Then try to bundle. It dies with an error.

When trying to update a certain gem (bundle update gemname) I get the following error message and the gem is not updated:

Error loading RubyGems plugin "/home/capistrano/.rvm/gems/ruby-2.0.0-p195@global/gems/rubygems-bundler-1.2.0/lib/rubygems_plugin.rb": Could not find 'rubygems-bundler' (>= 0) among 270 total gem(s) (Gem::LoadError)

This https://gist.github.com/mpapis/33668e76396ce52f5612#file-rubygems_plugin-rb fixed the error message but the gem still is not updated. Bundler is just running trough