Issue #16776 has been updated by mame (Yusuke Endoh).


Eregon (Benoit Daloze) wrote in #note-5:
> Maybe coverage could use a second type of internal tracepoint to fix this=
 issue (internal coverage handlers are likely pure C and so don't emit any =
extra event)?

Yes, it is the plan B that I have.  I don't like it as it is very ad-hoc, t=
hough.  If @ko1 rejects the priority, I'll give it a try to implement.

----------------------------------------
Bug #16776: Regression in coverage library
https://bugs.ruby-lang.org/issues/16776#change-85085

* Author: deivid (David Rodr=EDguez)
* Status: Open
* Priority: Normal
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN
----------------------------------------
Hi!

I noticed a regression in the coverage library. I tried to write a minimal =
program to show it, hopefully it gives some clues or where the issue might =
lie.

In ruby 2.5.8 and earlier, the following program would print `{:lines=3D>[1=
, 1, nil]}`, showing that the body of the "foo" method was run once. Howeve=
r, on newer rubies, it prints `{:lines=3D>[1, 0, nil]}`, which is incorrect=
 because the "foo" method body has actually been run once.

This is the repro script:

```ruby
# frozen_string_literal: true

require "coverage"
Coverage.start(lines: true)

code =3D <<~RUBY
  def foo
    "LOL"
  end
RUBY

File.open("foo.rb", "w") { |f| f.write(code) }

require_relative "foo"

TracePoint.new(:line) do |_tp|
  foo
end.enable do
  sleep 0
end

res =3D Coverage.result
puts res[File.expand_path("foo.rb")]
```




-- =

https://bugs.ruby-lang.org/

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=3Dunsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>