Module: SimpleCov::Production

Defined in:
lib/simplecov/production.rb,
lib/simplecov/production/error.rb,
lib/simplecov/production/file_sink.rb,
sig/simplecov.rbs,
sig/simplecov.rbs

Defined Under Namespace

Modules: _Sink Classes: Error, FileSink

Class Method Summary collapse

Class Method Details

.flush ⇒ Boolean

Called by the flush thread every interval; public so a deploy hook (or a signal handler) can force one. An empty delta counts as accepted.

Returns:

  • (Boolean) 


48
49
50
51
52
53
54
55
 
# File 'lib/simplecov/production.rb', line 48

def flush
  return false unless running?

  @mutex.synchronize do
    pull_pending
    push_pending
  end
end

.reset! ⇒ void

This method returns an undefined value.

registration flag survives on purpose: one hook per process.



73
74
75
76
77
78
79
80
81
 
# File 'lib/simplecov/production.rb', line 73

def reset!
  @running = false
  wake_flush_thread if @thread
  @thread&.join
  @thread = nil
  @pending = {}
  @pid = nil
  @started_coverage = false
end

.running? ⇒ Boolean

Both are small Integers, so identity settles it: a measurement belongs to the process that started it and to no other.

Returns:

  • (Boolean) 


42
43
44
 
# File 'lib/simplecov/production.rb', line 42

def running?
  !!@running && @pid.equal?(Process.pid)
end

.start(root: Dir.pwd, sink: nil, flush_interval: 60, flush_jitter: nil, sample_rate: 1.0, max_buffered_lines: 1_000_000) ⇒ Object

Answers false, with a warning naming why, when it safely declines (already running, another Coverage owner such as a test suite, or a runtime without oneshot lines). Configuration mistakes raise Error instead: a typo'd interval should fail deployment, not quietly measure wrong.

In a forking server, call this from the worker-boot hook (Puma's on_worker_boot, Unicorn's after_fork): the flush thread does not survive a fork, and start in the child picks the inherited measurement back up.



25
26
27
28
29
30
31
32
33
34
35
36
37
38
 
# File 'lib/simplecov/production.rb', line 25

def start(root: Dir.pwd, sink: nil, flush_interval: 60, flush_jitter: nil, sample_rate: 1.0,
  max_buffered_lines: 1_000_000)
  validate!(flush_interval, flush_jitter, sample_rate, max_buffered_lines)
  return warn_decline("SimpleCov::Production is already running") if running?
  return false unless claim_coverage

  configure(root, sink, flush_interval: flush_interval, flush_jitter: flush_jitter,
    sample_rate: sample_rate, max_buffered_lines: max_buffered_lines)
  @running = true
  @pid = Process.pid
  install_at_exit
  spawn_flush_thread
  true
end

.stop ⇒ Boolean

Returns:

  • (Boolean) 


57
58
59
60
61
62
63
64
65
66
67
68
69
 
# File 'lib/simplecov/production.rb', line 57

def stop
  return false unless running?

  @running = false
  wake_flush_thread
  @thread&.join
  @mutex.synchronize do
    pull_pending
    push_pending
    Coverage.result(stop: true, clear: true)
  end
  true
end