CommandSet shouldn't spill its internal state.

[ghost]

@commands is the object Pry::CommandSet uses to keep track of its commands. We expose it for everyone at Pry::CommandSet#commands but we shouldn't.

Pry::CommandSet is Enumerable already and anything that mutates @commands should be a method on Pry::CommandSet. exposing also makes this weird API possible:

Pry.commands.commands.each do
  # WAT?
end

Pry.commands.each do 
  # ok.
end

[ghost]

There's also this:

Pry.commands.commands["help"] = "hahaha!"
Pry.commands["help"] = "hahaha!" # TypeError.

[ghost]

Pry.commands.commands["help"] = "hahaha!" also totally fucks pry. ^D is the only way to fix it/exit.
CommandSet#[]= tries to stop you shooting yourself in the foot.

[kyrylo]

Fixed by #979.

[ghost]

@kyrylo that PR didn't fix this.

[kyrylo]

Sorry, my bad!

[yui-knk]

Now

[12] pry(main)> Pry.commands.commands["help"]
NoMethodError: undefined method `commands' for #<Pry::CommandSet:0x007fe2d4194dd0>
from (pry):11:in `__pry__'

[15] pry(main)> Pry.commands["help"]  = "hhh"
TypeError: command is not a subclass of Pry::Command
from pry/lib/pry/command_set.rb:346:in `[]='

Maybe this issue is fixed ?

[kdaigle]

On, pry (0.9.12.6), I get the following while running the same thing @yui-knk did:

[2] pry(main)> Pry.commands.commands["help"]
=> Pry::Command::Help
[3] pry(main)> Pry.commands["help"] = "hhh"
NoMethodError: undefined method `[]=' for #<Pry::CommandSet:0x00000109d2bc70>
from (pry):3:in `<main>'
[4] pry(main)>

But, I get the same response while on master. So it looks like this is fixed on master:

[1] pry(main)> Pry.commands.commands["help"]
NoMethodError: undefined method `commands' for #<Pry::CommandSet:0x00000104272428>
from (pry):1:in `<main>'
[2] pry(main)> Pry.commands["help"] = "hhh"
TypeError: command is not a subclass of Pry::Command
from /opt/rubies/2.0.0/lib/ruby/gems/2.0.0/bundler/gems/pry-ce952527cd5f/lib/pry/command_set.rb:346:in `[]='
[3] pry(main)>

It looks like c05c8f3 and 77a9c00 by @johnny5- fixed this. 👍

[Ferdy89]

@ghost @kyrylo @yui-knk as @kdaigle mentioned, this looks like it has been solved for a while. Could you close the issue?