Feature #19094: `sleep(nil)` vs `sleep()` and replicating the default implementation.

• Status: Closed
• Assignee: ioquatix (Samuel Williams)
• Target version: -

[ioquatix (Samuel Williams)]

> sleep(nil)
(irb):1:in `sleep': can't convert NilClass into time interval (TypeError)

However, I feel that this makes implementing a compatible sleep method a little difficult.

def sleep(time = nil)
  if time
    sleep_some(time)
  else
    sleep_forever
  end
end

Can we consider allowing sleep(nil) and sleep(false) to be the same as sleep() to simplify this behaviour?

Otherwise, it's hard to proxy (e.g. fiber scheduler, or even just a plain old variable argument).

e.g.

class Sleeper
  def initialize(time = nil)
    @time = time
  end

  def sleep
    Kernel::sleep(@time) # Hard to get the behaviour of `sleep()` here without explicitly handling/checking.
  end
end

[jeremyevans0 (Jeremy Evans)]

• Tracker changed from Bug to Feature
• Backport deleted (2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN)

[ioquatix (Samuel Williams)]

• Tracker changed from Feature to Bug
• Description updated (diff)
• Backport set to 2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN

[ioquatix (Samuel Williams)]

• Tracker changed from Bug to Feature
• Backport deleted (2.7: UNKNOWN, 3.0: UNKNOWN, 3.1: UNKNOWN)

[sawa (Tsuyoshi Sawada)]

I don't think it is hard to implement. All you need is a twinkle little star.

class Sleeper
  def initialize(time = nil)
    @time = time
  end

  def sleep
    Kernel::sleep(*@time)
  end
end

[ioquatix (Samuel Williams)]

Haha, great, but I think it's odd you can't just write sleep(nil). I guess Ruby uses Qundef internally, but user can't write sleep(undef), but sleep(*nil) is close. Actually, I didn't even know you can write sleep(*nil) to get that behaviour, so I think it's not a good solution in general, it's too obscure.

[ioquatix (Samuel Williams)]

By the way, my motivation was to make this easier for the user: https://github.com/socketry/async/issues/161

[Eregon (Benoit Daloze)]

Mutex#sleep accepts an explicit nil argument and it's documented.
+1, this inconsistency is rather weird (and rather messy to implement in TruffleRuby).

I don't know about false though, that's can't convert FalseClass into time interval (TypeError) for both sleep methods which seems fair.

[matz (Yukihiro Matsumoto)]

I understand the usefulness of the proposal. Accepted.
But do not propose other changes just for consistency with this change.

Matz.

[ioquatix (Samuel Williams)]

PR: https://github.com/ruby/ruby/pull/7484

[ioquatix (Samuel Williams)]

• Status changed from Open to Closed

Merged.