Misc #21348: Should Tracepoint track retry as another "call" event?

• Status: Open
• Assignee: -

[karthikc (Karthik Chandrasekariah)]

When retry is executed in a method, Tracepoint records it as a new "call" event.

# tracepoint-retry.rb

# method that retries once
def foo
  attempts ||= 1
  raise "Fail" if attempts == 1
rescue
  attempts += 1
  retry
end

trace = TracePoint.new(:call, :return) do |tp|
  p [tp.event, tp.method_id]
end

trace.enable
foo

$ ruby -v
ruby 3.4.2 (2025-02-15 revision d2930f8e7a) +PRISM [x86_64-darwin24]

$ ruby tracepoint-retry.rb
[:call, :foo]
[:call, :foo]
[:return, :foo]

It results in multiple "call" events and a single "return" event. Since the retry doesn't technically leave and re-enter the method, should we change Tracepoint to not fire the second "call" event?

Context - I am building a library that tracks everything that happened in a block of code. This behavior makes it look like foo was called within foo but the outer call never returned/completed.

[Eregon (Benoit Daloze)]

I agree with the OP that :call shouldn't trigger a second time here, since there is only one call to foo.
retry is similar to a loop, and of course we don't add extra :call TracePoint events for loops.