on
Debugging in Rails with Pry, `bundle open` and `puts`
This week I got to dig into some debugging that I really enjoyed. Well, if you had asked me in the middle of it I might not have been having fun, but I found a solution! I used a handful of tools to approach debugging and they all gave me a little bit more information to solve the problem that I didn’t have before.
Here is a bit about each tool and what I find useful about it. Let me know if you have other Ruby or Rails debugging tips!
Pry!
Pry is an amazing tool for Ruby. It gives you all kinds of tools like ? my_method to show you documentation or ls MyObject to show you all the methods you can call on an object. You can even use cd to move into an object!
Here is a tiny class with a pry binding in it to show you how a few of the methods work.
class User
def initialize(name)
@name = name
end
def name
binding.pry
@name
end
end
I started up from the command line and initialized a new user. When I call the name method, Pry prints out the lines around the binding and gives you an interactive prompt.
[1] pry(main)> me = User.new('cassidy')
=> #<User:0x00007fd807ed87a8 @name="cassidy">
[2] pry(main)> me.name
From: (pry) @ line 9 User#name:
7: def name
8: binding.pry
=> 9: @name
10: end
[1] pry(#<User>)>
At this point I could type @name to see what the value of my instance variable is.
[1] pry(#<User>)> @name
=> "cassidy"
Ah good, it’s what I expected. Let’s use step to move the execution forward one more frame.
[2] pry(#<User>)> step
From: /Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/pry-0.12.2/lib/pry/pry_instance.rb @ line 388 Pry#evaluate_ruby:
383: def evaluate_ruby(code)
384: inject_sticky_locals!
385: exec_hook :before_eval, code, self
386:
387: result = current_binding.eval(code, Pry.eval_path, Pry.current_line)
=> 388: set_last_result(result, code)
389: ensure
390: update_input_history(code)
391: exec_hook :after_eval, result, self
392: end
[1] pry(#<Pry>)>
Neat! It looks like the next frame is Pry itself! Ever wonder how pry works? Check out all the methods that are here!
[1] pry(#<Pry>)> ls
Pry#methods:
add_sticky_local commands= eval extra_sticky_locals= last_dir memory_size print quiet? should_print?
backtrace complete eval_string hooks last_dir= memory_size= print= raise_up show_result
backtrace= config eval_string= hooks= last_exception output process_command raise_up! sticky_locals
binding_stack current_binding evaluate_ruby inject_local last_exception= output= process_command_safely raise_up_common suppress_output
binding_stack= current_context exception_handler inject_sticky_locals! last_file output_array prompt repl suppress_output=
color custom_completions exception_handler= input last_file= output_ring prompt= reset_eval_string update_input_history
color= custom_completions= exec_hook input= last_result pager push_binding run_command
command_state editor exit_value input_array last_result= pager= push_initial_binding select_prompt
commands editor= extra_sticky_locals input_ring last_result_is_exception? pop_prompt push_prompt set_last_result
instance variables:
@backtrace @command_state @custom_completions @indent @last_exception @last_result_is_exception @prompt_stack @suppress_output
@binding_stack @config @eval_string @input_ring @last_result @output_ring @stopped
locals: _ __ _dir_ _ex_ _file_ _in_ _out_ _pry_ code result
I’m not going to dig around in there too much, but lets see what next does.
From: /Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/pry-0.12.2/lib/pry/pry_instance.rb @ line 390 Pry#evaluate_ruby:
383: def evaluate_ruby(code)
384: inject_sticky_locals!
385: exec_hook :before_eval, code, self
386:
387: result = current_binding.eval(code, Pry.eval_path, Pry.current_line)
388: set_last_result(result, code)
389: ensure
=> 390: update_input_history(code)
391: exec_hook :after_eval, result, self
392: end
It moved us one line forward in the current frame. So behind the scenes, it executed all of line 38 and returned us to this same stack frame because there was more code to execute here. Neat!
Okay, enough fiddling around with pry. Let’s type continue and we’ll finally see the result of the name method.
[1] pry(#<Pry>)> continue
=> "cassidy"
Bundle Open
What happens when you start stepping through a call stack in pry and you find yourself wandering around stack frames that belong to a Ruby Gem you’re using. When this happens to me I like to go checkout what the code actually looks like in that gem with bundle open.
Here’s the error I was working on today. You’ll see that the error is raised in a third party library. I might have made a mistake that caused it, but my mistake is so far down the call stack that I don’t see it!
2019-08-01T21:42:43.628Z 34126 TID-ower3a8uc WARN: NoMethodError: undefined method `body' for nil:NilClass
2019-08-01T21:42:43.628Z 34126 TID-ower3a8uc WARN: /Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/carrierwave-0.10.0/lib/carrierwave/storage/fog.rb:228:in `read'
/Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/carrierwave-0.10.0/lib/carrierwave/uploader/cache.rb:77:in `sanitized_file'
/Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/carrierwave-0.10.0/lib/carrierwave/uploader/cache.rb:116:in `cache!'
/Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/carrierwave-0.10.0/lib/carrierwave/uploader/versions.rb:226:in `recreate_versions!'
/Users/cassidy/.rbenv/versions/2.3.8/lib/ruby/gems/2.3.0/gems/carrierwave_backgrounder-0.4.2/lib/backgrounder/workers/process_asset.rb:12:in `perform'`
I had no idea what was nil here that shouldn’t be so I ran bundle open carrierwave to open the gem’s codebase in a text editor. Once I was there, I found the line where the error was raised and put a binding.pry in it so I could get some better insight as to what was going on.
puts!!!
Sometimes pry isn’t going to work for you. This is especially true if you’re using a multi-threaded web server or Sidekiq to process background jobs. When this happens, you’ll get an interactive prompt for a few seconds then you’ll get booted out because another process is trying to write logs to the same master process. Argh! This can be frustrating.
My favourite thing to do here is to print my own logs. I try to make these stand out so I don’t miss them when the logs are going by. Here’s some pseudocode that resembles the problem I was solving today.
def filename
puts '---------------------'
puts 'in filename'
puts model.attributes['image']
puts model.persisted?
puts model.image_changed?
puts '---------------------'
return model.attributes['image'] if model.persisted? && !model.image_changed?
new_filename = "#{SecureRandom.hex}.#{file&.extension}"
puts new_filename
new_filename
end
This gave me some clear output so I could see what was happening each time this method was called. The great thing about puts debugging is that you’re never wondering if you lost count of which iteration you’re on when a method gets called multiple times.
Conclusion
When I was working on my problem today, I used all three of these tools together. Pry is my go-to, but when that isn’t enough, I’ll reach for something else. I ended up learning lots about the CarrierWave life cycle that will make the rest of my task so much easier! I even learned about some methods CarrierWave has that aren’t documented.