LazyRegister concurrent register bug

[kvokka]

If yout take a look at

concurrent-ruby/lib-edge/concurrent/lazy_register.rb

Line 61 in bbeacbc

def register(key, &block)

you will see, that it has a bug.

When block evaluation time is stochastic, then even 2 register calls will produce unpredictable result.

* `concurrent-ruby` version:       master
* `concurrent-ruby-edge` used:     yes

Will be happy to fix it, if u don't mind.

[pitr-ch]

Thank you for you interest. However, I don't see the problem, could you please elaborate. Thanks.

[kvokka]

Thank you for such a quick response.

After spending some time, I realized, that the is no a word about concurrent write of user data, only the registry itself. So, the developer should to manage it. This means, that for my case I should create a subclass, which allows the concurrent immutable access to the data.

Will do it anyway for my needs. Is it good idea to add it as a PR to -edge afterwards?

[pitr-ch]

What is your use-case?
Yeah keep the new class next to the lazy_reqister in edge for now please. I would recommend to submit a draft PR so we can discuss early. Thanks!