I've written a data collection script for Cacti in Ruby and it runs fine from the command line but Cacti runs the script via "env -i" which strips the environment so Ruby can't find the rubygems library ("in `require': no such file to load -- rubygems (LoadError)"). How might I work around this?
Can you modify any of the following in your Ruby script: $:, $-I or $LOAD_PATH? These all just point to the same array which specifies where Ruby looks for classes and other ephemera...
>> $LOAD_PATH
=> ["/usr/local/lib/ruby/site_ruby/1.8", "/usr/local/lib/ruby/site_ruby/1.8/i686-darwin9.5.0", "/usr/local/lib/ruby/site_ruby", "/usr/local/lib/ruby/1.8", "/usr/local/lib/ruby/1.8/i686-darwin9.5.0", "."]
#!/bin/sh
#export LOAD_PATH=whatever
#export RUBYLIB=whatever
#export RUBYOPT=whatever
#export RUBYPATH=whatever
#export RUBYSHELL=whatever
#export PATH=$PATH:whatever
exec ruby -x. $0 "$@"
#!/usr/bin/ruby
require 'rubygems'
require 'open4' # or whatever
# rest of ruby script here
This is a shell script that runs ruby with -x, which will cause the interpreter to skip lines until it finds #!.*ruby
. This will give you a chance to restore the environment. The .
after -x
is a noop, you can take out the .
, or replace it with a directory. Ruby will cd there before running the script.
I'm actually guessing that this is not really what you want, since this could have been done without any trickery by just making two scripts, one for the shell, one for Ruby. Perhaps the list of environment variables Ruby cares about will help...
I don't think $LOAD_PATH
used for gems (at least, not exclusively). You might want to look at a couple environment variables that haven't been mentioned here yet:
ENV['GEM_HOME']
ENV['GEM_PATH']
You can see your current paths for gems with:
require 'rubygems'
puts Gem.path
A partial answer might be here: comp.lang.ruby post