remove cli.rb to prevent loading of plugin

[mikz]

pry will load all cli.rb files from plugins, no matter when they are disabled or not

and loading cli.rb will load all extensions and fully loading the plugin
so Pry.plugins['byebug'].disable! does not work
neither does Pry.config.should_load_plugins = false

removing cli.rb fixes it
I need to disable pry-byebug for some cases because of #69.

looks like it was introduced (https://github.com/deivid-rodriguez/pry-byebug/commits/77e1c046f89ae4b4dc255e6c5867415a20a78cec/lib/pry-debugger/cli.rb) to support pry-remote but that does not work much (#33)

Other pry plugins using cli.rb actually do something with the Pry::CLI which is not the case here. See https://github.com/pry/pry-exception_explorer/blob/540d7c9d139d6493083c3dc5c18315c1db100cc1/lib/pry-exception_explorer/cli.rb for an example.

[deivid-rodriguez]

Hi @mikz, thanks for this.

I think we need some more research to get this merged. Please refer to this comment for context and let me know!

[mikz]

@deivid-rodriguez I see. So you need pry-byebug to monkeypatch Pry.start and the only place to do that is cli.rb ? Because everything else is loaded afterwards ? Uf.

Would any of the Pry hooks be helpful there?

[deivid-rodriguez]

I honestly don't know why this is needed, but removing it seems to cause a regression..., at least it did in the past 😞

So I prefer to leave it for now unless someone can make sure we don't regress.

[mikz]

I can make sure that it does not regress. I just need to know why it tries to monkeypatch Pry.start. Would be enough just initializing the Byebug::PryProcessor when the plugin is initialized in https://github.com/pry/pry/blob/f5c97cac103a9f04b854ed5b117a8cc27c7e8dd6/lib/pry/pry_class.rb#L177 ?

[deivid-rodriguez]

@mikz If you can get pry-byebug working without monkeypatching Pry.start, I think that's a further improvement besides this PR! 👍 I don't remember any technical reason why monkeypatching would be compulsory. It's just the way it's always been implemented but I'm up for changing it.

[mikz]

Ok. To drop the cli.rb I'd need to get rid of the monkeypatch too. From what I see it should be just fine.

I'll try this to verify the regression:

echo "require 'pry'; binding.pry; puts 'next statement'" > foo.rb
ruby foo.rb
# next

[deivid-rodriguez]

Ideally we should add a regression test for posterity. But if you don't have time I guess we could live without it.

[deivid-rodriguez]

@mikz Ping. Did you forget about this or just decided not to work on it?

[mikz]

@deivid-rodriguez kind of both. IIRC there was more places to fix and I just did not have enough time.

I'd like to fix this in the future, just not sure when I'll find time.

[deivid-rodriguez]

Sounds good to me!