# Plugin type discovery/registration - How do I discover what

**URL:** <https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278>\
**Category:** Ruby\
**Created:** [February 27, 2006, 6:53pm UTC](https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278 "2006-02-27T18:53:22Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![jbell38](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/edb3f5/32.png) [@jbell38](https://www.ruby-forum.com/u/jbell38)\
**Post date:** [February 27, 2006, 6:53pm UTC](https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278/1 "2006-02-27T18:53:22Z")

</div>

In my current project I am in need of a standard plugin mechanism where  
I  
discover what plugins are available by what exists in a directory. I was  
able to easily simply ‘require’ these ruby files in. However I want to  
take  
it a step further and know (discover) what classes were in these files  
including the module name so that I can interrogate them further and  
potentially create them and call an initialize type method which might  
be  
used to register yourself and or do other startup processing.

Is there an easy way to know what classes were loaded during the require  
or  
is there a better way to do this sort of thing? I think the Rails plugin  
mechanism works similarly to what I am looking for in that it loads  
plugins  
from a directory and invokes an initialize method on each so they can do  
their own registration/configuration work.

PS. It would also be nice if the plugins can use whatever module space  
they  
want so there won’t be class conflicts, so I would prefer suggestions  
and  
ideas that might allow this functionality, so not only do I get the  
class  
name but the actual module::class name so that I could create the proper  
object using this information.

Thanks for any suggestions or help!

Jeff

---

<div class="post-metadata">

**Author:** ![jbell38](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/edb3f5/32.png) [@jbell38](https://www.ruby-forum.com/u/jbell38)\
**Post date:** [February 27, 2006, 7:05pm UTC](https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278/2 "2006-02-27T19:05:59Z")

</div>

On Feb 27, 2006, at 11:52 AM, Jeff B. wrote:

> Is there an easy way to know what classes were loaded during the  
> require or  
> is there a better way to do this sort of thing?

See if this recent post helps:

[http://ruby-talk.org/cgi-bin/scat.rb/ruby/ruby-talk/181245](http://ruby-talk.org/cgi-bin/scat.rb/ruby/ruby-talk/181245)

Also see Ara’s dynaload library:

[http://rubyforge.org/frs/?group\_id=1024&release\_id=3786](http://rubyforge.org/frs/?group_id=1024&release_id=3786)

Hope that helps.

James Edward G. II

---

<div class="post-metadata">

**Author:** ![jbell38](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/edb3f5/32.png) [@jbell38](https://www.ruby-forum.com/u/jbell38)\
**Post date:** [February 27, 2006, 7:21pm UTC](https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278/3 "2006-02-27T19:21:17Z")

</div>

James Edward G. II wrote:

> Also see Ara’s dynaload library:
> 
> [http://rubyforge.org/frs/?group\_id=1024&release\_id=3786](http://rubyforge.org/frs/?group_id=1024&release_id=3786)

this is relevant too, I think  
[http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/180017](http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/180017)

---

<div class="post-metadata">

**Author:** ![jbell38](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/edb3f5/32.png) [@jbell38](https://www.ruby-forum.com/u/jbell38)\
**Post date:** [February 27, 2006, 8:07pm UTC](https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278/4 "2006-02-27T20:07:21Z")

</div>

Jeff B. wrote:

> mechanism works similarly to what I am looking for in that it loads plugins  
> from a directory and invokes an initialize method on each so they can do  
> their own registration/configuration work.
> 
> PS. It would also be nice if the plugins can use whatever module space they  
> want so there won’t be class conflicts, so I would prefer suggestions and  
> ideas that might allow this functionality, so not only do I get the class  
> name but the actual module::class name so that I could create the proper  
> object using this information.
> 
> Thanks for any suggestions or help!

Kernel#module\_eval is a good tool for wrapping up plug-ins inside a new  
namespace. I wrote the “script” library on raa to simplify doing that.  
For example:

$ cat plugins/plug1.rb  
class Plug1  
end

K = 42

$ cat plugins/plug2.rb  
class Plug2  
end

K = 43

$ cat plugin.rb  
require ‘script’

plugins = Dir.glob(“plugins/\*.rb”).map do |file|  
Script.load file  
end

p plugins

plugins.each do |plugin|  
puts “plugin #{plugin} defines these constants:”  
plugin.constants.each do |const\_name|  
val = plugin.const\_get(const\_name)  
puts " %20s = %s" % [const\_name, val.inspect]  
end  
puts  
end

$ ruby plugin.rb  
[#Script:/path/to/plug1.rb, #Script:/path/to/plug2.rb]  
plugin #Script:/path/to/plug1.rb defines these constants:  
Plug1 = #Script:0xb7d61414::Plug1  
K = 42

plugin #Script:/path/to/plug2.rb defines these constants:  
Plug2 = #Script:0xb7d613c4::Plug2  
K = 43

(I substituted /path/to/ for the full path to save space.)

Of course, you can filter out only the classes, and skip other  
constants.

---

<div class="post-metadata">

**Author:** ![jbell38](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/edb3f5/32.png) [@jbell38](https://www.ruby-forum.com/u/jbell38)\
**Post date:** [February 27, 2006, 10:20pm UTC](https://www.ruby-forum.com/t/plugin-type-discovery-registration-how-do-i-discover-what/74278/5 "2006-02-27T22:20:25Z")

</div>

Thanks for the quick responses and wonderful suggestions, these are  
great! I  
can see where each of these techniques would be useful in slightly  
different  
ways.