Configuring Pry

Configuring Pry

I use Pry as my REPL for all my Ruby projects. In fact, Pry is one of the reasons Ruby has “stuck” with me. I find the commands both intuitive and helpful. I have a text expansion that I constantly use in development: ;pry -> binding.pry if !@jsw_skip_pry The if clause is there to allow easy exits from loops - once I see what I need to see, I set the instance variable to true to avoid all subsequent breaks.

I have been using Pry for years and I have never messed with its configuration. Just this week I was debugging a pretty involved class. The instantiated class often has A LOT of instance variables (maybe that’s a code smell? That’s a topic for another day which may never come). I was getting an exception in this instantiated class unrelated to the instance variables. Pry’s default exception handler, which is one of the reason’s I love Pry, in this case was spitting out all of the instance variables which was taking up literally hundreds of lines in my terminal and I was having to scroll through all of these lines to see the actual exception that was being thrown. It got to be too much. I went in search of options to tame the exception output. Pry did not disapoint. I found a great resource for configuring Pry’s exception handler. Lot’s of options there. The big takeaways for me were that it’s fairly trivial to override its default handler and that you can do it in either the active session, or in a .pryrc file¹.

Here is my Pry exception handler from my .pryrc in my home directory:

Pry.config.exception_handler = proc do |output, exception, _|
  output.puts "#{exception.class.name}: #{exception.message}\n"
  exception.backtrace.each {|ebt| output.puts ebt if ebt.match?(/^(?!.*(__pry__|\/gems\/|<main>)).*$/)}
end

This handler omits the default handler’s object inspection, which is what causes all the instance variables to be displayed. I am also filtering the backtrace to exclude any lines referencing, pry, gems, and main because 99% of the time I am only interested in errors from my code and this dramtically reduces the backtrace. If I ever need the more verbose default exception, I can always comment out my code in my ~/.pryrc and start a new session, or, better yet, create a different handler in my current Pry session:

pry_instance.config.exception_handler = proc do |output, exception, _|
  output.puts "#{exception.class}: #{exception.message}"
  output.puts "from #{exception.backtrace.first}"
end

1. .pryrc files follow a common configuration convention in that your directory tree is searched from root to your current directory for .pryrc files and options are loaded from any found. Settings from earlier .pryrcs found closer to your working directory override the earlier setting. ↩