From: cldwalker <gabriel.horner@...>
Date: 2009-05-14T00:50:04+09:00
Subject: Re: how do you create a large number of irb commands?

On May 13, 11:07 am, ghorner <gabriel.hor...@gmail.com> wrote:
> I'm wondering what you all consider the best way for creating a large
> number of irb commands. By commands I'm referring to methods for the
> default irb object like irb,fg and jobs. Here are some of the
> different ways that come to mind:
>
> * Extend the default irb object with modules.
> * Include modules into the default irb object. However this has the
> nasty side effect of making those methods accessible to all objects.
> * Just define methods in the default irb object. Simple and seems to
> be the common way in irbrc's.
> * Use irb's internal command system i.e. self.def_extend_command() inhttp://github.com/rubyspec/matzruby/blob/9ca4e66f27e2563512afdd4c70af...
> Seems like overkill.
>
> To me the only serious options are extending or just defining the
> method. By extending I get an ancestry trace on where commands came
> from which is important when dealing with a large number of commands.
> Also, if i do switch subsessions or workspaces I can just extend in
> needed functionality. Since defining methods in the irb object lacks
> the ancestry trace and the ability to reuse them in other context, my
> vote is for extending.
>
> Thoughts?
>
> Gabriel

I forgot to mention the option of commands as private methods in
Kernel space. Like the include option, I think it's too dirty as it
gives all objects access to those methods through send(). Although
this doesn't matter for a couple of commands, it could lead to some
dangerous side effects for many commands. My vote is still with
extending modules.

Gabriel