16:41 <aweis> hello, I am trying to deploy bundler to a large rails deployment with multiple machines already in production

16:42 <aweis> they use the system gems for all gems in the system, and I have gotten everything to work fine for scripts that load the rails environment, as there is a .bundle/config that loads the gems from vendor/bundle

16:43 <aweis> however, there are hundreds of other scripts on the system, that do not load the rails environment. How do I get those scripts to load gems from vendor/bundle?

16:45 <aweis> I have tried bundle installing the Gemfile into the system gems directory, but this does not work for gems from :git => ... (as is pointed out in the documentation) and I was wondering what the best practices are for these situations

16:47 <jhass> aweis: either run them through bundle exec or add require "bundler/setup" as the first require to them

16:49 <aweis> jhass: this requires the need to modify multiple scripts? I have found a way around changing all the scripts by adding a ruby shim, where when calling ruby, I am actually already requiring bundler setup exec /usr/bin/ruby1.9.3 -r 'rubygems' -r 'bundler/setup'

16:50 <jhass> -r 'rubygems' is implicitly done by ruby for you since 1.9 (which is EOL'd btw), but yeah, if neither editing the scripts nor all invocations is an option that's probably your way to go

17:12 <aweis> jhass: thanks for confirming the approach as valid. It is very difficult for me to get a good idea on how to retrofit old systems into using bundler, and its useful to have someone to talk over the problems with.

17:13 <jhass> I'd still prefer one of the other two approaches though

17:13 <jhass> it is a workaround

17:15 <indirect> aweis: jhass: `bundle exec` is effectively identical to `ruby -rbundler/setup`

17:16 <jhass> yes, but I think it transports intent a bit better

17:16 <indirect> aweis: I strongly recommend upgrading from Ruby 1.9.3, since it will stop getting security patches soon (already has stopped, maybe)

17:16 <jhass> also we're talking about shimming ruby to bundle exec here

17:16 <indirect> you could also set the shell environment variable `RUBYOPT=-rbundler/setup`

17:16 <indirect> which effectively forces all ruby processes to be bundle-exec'ed

17:17 <jhass> ruby 1.9.3 will no longer receive any official security patches, Ruby 2.0 will cease to receive them in late February 2016

17:17 <indirect> aweis: you can install the bundle into system gems, but you still need to use bundler, either via `bundle exec` or via `require 'bundler/setup'` or with `RUBYOPT=-rbundler/setup`

17:18 <indirect> there are many more problems besides just git gems that require it

17:27 <aweis> indirect: my major issue is that the system gem directory is not writable, so the :git gems wont be stored there. I believe the :git gems would then get placed in the .gem directory, but the bundler docs highly suggest to avoid system gems. What other problems are your referring to?

17:28 <indirect> aweis: if the system gem directory is not writable, then Bundler cannot use it

17:28 <indirect> if you have any git gems at all, write access is required