Making SimpleCov work when executing gem binary - ruby

Question

I am trying to add SimpleCov coverage to my gem that has a binary.
I would like to test its command line interface, so I wish to test the binary execution itself, and not the library it uses.
I am getting no coverage report (0 LOC) by SimpleCov.
As I understand, the issue is most likely due to the fact that in my test (either cucumber features or rspec specs), I am executing the gem's binary with system or popen3, but I don't know how I can tell SimpleCov to "follow through" (or if I am barking at the right tree...).
I have tried playing with SimpleCov.command_name, SimpleCov.pid = $$, SimpleCov.track_files and almost every other remotely related configuration I found.
I do not wish to use Aruba, although I have tried reviewing their source to look for possible answers.
Related code snippets:
# spec_helper.rb
require 'simplecov'
SimpleCov.start
require 'rubygems'
require 'bundler'
Bundler.require :default, :development
# test_spec.rb
require 'spec_helper'
describe "my bin" do
it "should be covered" do
system 'bin/runme'
end
end
I have prepared a minimal repo as an easy testing ground, if that helps at all.

Answer 1

well it's not easy to find a solution to this issue in particular.
not sure this solves your problem but maybe a start?
without the start block which modifies the filters i could not get SimpleCov to watch the bin directory (using the sample github repo you provided).
Used command_name to give the main processes coverage reports a name and then in the fork used command_name to give the forked processes report a name (SimpleCov merges them for us as long as they have different names).
then used load to load the bin file instead of using system.
(I couldn't figure out a way to make system or spawn add to the coverage reports, maybe if you called it through a script that restarts SimpleCov for you with an alternate command_name)
again, not sure if this is exactly what you are looking for but may be a start. code below:
# spec_helper.rb
require 'simplecov'
SimpleCov.command_name "main_report"
SimpleCov.start do
filters.clear # This will remove the :root_filter and :bundler_filter that come via simplecov's defaults
add_filter do |src|
!(src.filename =~ /^#{SimpleCov.root}/) unless src.filename =~ /bin/ #make sure the bin directory is allowed
end
end
require 'rubygems'
require 'bundler'
Bundler.require :default, :development
# test_spec.rb
require 'spec_helper'
describe "my bin" do
it "should be covered" do
pid = Process.fork do
SimpleCov.start do
command_name "bin_report_section"
end
load "bin/runme"
end
end
end
result:
Coverage report generated for bin_report_section, main_report to
/home/korreyd/simplecov-debug/coverage.
1 / 1 LOC (100.0%) covered.

Answer 2

Have you tried this? https://blog.yossarian.net/2018/04/01/Code-coverage-with-Simplecov-across-multiple-processes
Basically, in your spec_helper.rb
if ENV["COVERAGE"]
require "simplecov"
# Only necessary if your tests *might* take longer than the default merge
# timeout, which is 10 minutes (600s).
SimpleCov.merge_timeout(1200)
# Store our original (pre-fork) pid, so that we only call `format!`
# in our exit handler if we're in the original parent.
pid = Process.pid
SimpleCov.at_exit do
SimpleCov.result.format! if Process.pid == pid
end
# Start SimpleCov as usual.
SimpleCov.start
end
Then inside your bin/runme, add:
if ENV["COVERAGE"]
# Give our new forked process a unique command name, to prevent problems
# when merging coverage results.
SimpleCov.command_name SecureRandom.uuid
SimpleCov.start
end
Child process' test coverage will merge into the parent's process.
If you use SimpleCov's coverage_dir, make sure it's in all SimpleCov.start blocks so that results are written to same location.