# Ruby/LDAP on Rails **URL:** \ **Category:** Rails\ **Created:** [March 8, 2006, 11:21pm UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580 "2006-03-08T23:21:04Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![miracleofmagick](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/71c47a/32.png) [@miracleofmagick](https://www.ruby-forum.com/u/miracleofmagick)\ **Post date:** [March 8, 2006, 11:21pm UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580/1 "2006-03-08T23:21:04Z")
This is wierd. I’m trying to get Ruby/LDAP up and working, and at least the basics work fine under irb: irb(main):001:0\> require ‘ldap’ =\> true irb(main):002:0\> LDAP::Conn.new() =\> #LDAP::Conn:0x1b87ac irb(main):003:0\> So far, so good. All LDAP::Conn.new does is call ldap\_init with host and port parameters, which in turn just allocates the control structure – there’s no attempt to create a socket until later. When I switch to Rails, though, this gets strange. Running inside a brand new rails structure (this is 1.0.0), I do: $ script/console Loading development environment. > > require ‘ldap’ > > =\> true > > > LDAP::Conn.new() > > LDAP::ResultError: can’t initialise an LDAP session > > from (irb):2:in `initialize’ > > from (irb):2 > > > When I first saw this, it was on my mac `ruby -v` -\> ruby 1.8.4 (2005-12-24) [powerpc-darwin8.5.0] with ruby-ldap-0.9.2, and I’ve confirmed the behavior through ruby-ldap-0.9.4. This isn’t just an OSX thing, though. I see the exact same error on my gentoo box (originally 2004.1, but kept current, where `ruby -v` -\> ruby 1.8.4(2005-12-24) [i686-linux]), with everything up to and including ruby-ldap-0.9.2 installed through portage. Thanks! Eric ---
**Author:** ![miracleofmagick](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/71c47a/32.png) [@miracleofmagick](https://www.ruby-forum.com/u/miracleofmagick)\ **Post date:** [March 8, 2006, 11:53pm UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580/2 "2006-03-08T23:53:46Z")
On 3/8/06, Eric H. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > So far, so good. All LDAP::Conn.new does is call ldap\_init with host and > port parameters, which in turn just allocates the control structure – > there’s no attempt to create a socket until later. I don’t know anything more about LDAP::Conn than what you wrote here, but where do the host and port parameters come from? Is it possible that they aren’t accessible from within Rails? ---
**Author:** ![miracleofmagick](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/71c47a/32.png) [@miracleofmagick](https://www.ruby-forum.com/u/miracleofmagick)\ **Post date:** [March 8, 2006, 11:44pm UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580/3 "2006-03-08T23:44:35Z")
On 3/8/06, Eric H. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > =\> true > > > > LDAP::Conn.new() > > > LDAP::ResultError: can’t initialise an LDAP session > > > from (irb):2:in `initialize’ > > > from (irb):2 > > When I first saw this, it was on my mac `ruby -v` → ruby 1.8.4 > (2005-12-24) [powerpc-darwin8.5.0] with ruby-ldap-0.9.2, and I’ve > confirmed the behavior through ruby-ldap-0.9.4. Sorry, couldn’t duplicate it at work (RHEL): ``` ruby 1.8.2 (2004-12-25) [i386-linux] ruby-ldap-0.9.2 rails 1.0.0 Red Hat Enterprise 3 $ ./script/console Loading development environment. >> require 'ldap' => true >> LDAP::Conn.new() => # ``` If I find the time, I’ll try at home (Mac OS X). ---
**Author:** ![miracleofmagick](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/71c47a/32.png) [@miracleofmagick](https://www.ruby-forum.com/u/miracleofmagick)\ **Post date:** [March 10, 2006, 12:15am UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580/4 "2006-03-10T00:15:54Z")
Bah. Found a work around at least. It looks like this may be an issue with openldap 2.2.19 as installed in OSX 10.4.5. I build openldap 2.3.20 (without slapd, to save time), installed it under /usr/local, and pointed ruby-ldap at that for the build. It works like a champ now. Thanks to everybody who tried to help. -Eric ---
**Author:** ![miracleofmagick](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/71c47a/32.png) [@miracleofmagick](https://www.ruby-forum.com/u/miracleofmagick)\ **Post date:** [March 9, 2006, 12:19am UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580/5 "2006-03-09T00:19:43Z")
They’re both optional, and default to “localhost” and LDAP\_PORT (port 389, defined in ldap.c as `rb_define_const (rb_mLDAP, "LDAP_PORT", INT2NUM(389));` I’ve tried wrapping the actual ldap\_init call with a couple of printf()s, and can verify that it appears to be getting the arguments it expects. Josh on Rails wrote: > I don’t know anything more about LDAP::Conn than what you wrote here, > but > where do the host and port parameters come from? Is it possible that > they > aren’t accessible from within Rails? ---
**Author:** ![miracleofmagick](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/71c47a/32.png) [@miracleofmagick](https://www.ruby-forum.com/u/miracleofmagick)\ **Post date:** [April 10, 2006, 9:29pm UTC](https://www.ruby-forum.com/t/ruby-ldap-on-rails/54580/6 "2006-04-10T21:29:19Z")
In fact, it appears that I was wrong. The real problem was a shared lib problem between ruby-oci8 and ruby-ldap. If you have ruby-oci8 installed and linked against a 10g-series instantclient, then be aware that libclntsh.so has its own version of ldap\_connect() – and it doesn’t appear to work. Even if you aren’t using it, the ActiveRecord initialization sequence will load oci8lib.so and its dependencies. If you then have your ‘require “ldap”’ in environment.rb where it gets loaded later, then calls to ldap\_connect() will be handled by the oci version, not the openldap version. The workaround I found was to put ‘require “ldap”’ at the _top_ of environment.rb, prior to the require of ‘boot’. May this save someone a bit of time.