Nio forever #82
Nio forever #82
Conversation
This solves socketry/timers#58 , a bug that has been around. Long story short, timers were passing really short intervals on the actor, these were being passed to the reactor for select. Some of them were being cast to 0 before being passed to the nio.Channel#select method, which according to documentation, blocks forever when passed 0.
|
@tarcieri , patch release? I guess celluloid-io isn't usable without this, what do you think? |
|
@ioquatix , ping. |
|
I can take a look tonight... |
|
@tarcieri , please follow linked issue. I think that it's not nio4r anymore, but just old on that patch release for now. |
|
Before we merge this perhaps we should find out why nio4r is failing on travis. |
|
@ioquatix it's RuboCop. I can take care of that separately tonight, and pin it in the Gemfile so this doesn't happen again |
|
@tarcieri that would be great. Thanks. |
RuboCop 0.36 fixups
|
@TiagoCardoso1983 can you rebase? I fixed the RuboCop errors in #83 |
This solves socketry/timers#58 , a bug that has been around. Long story short, timers were passing really short intervals on the actor, these were being passed to the reactor for select. Some of them were being cast to 0 before being passed to the nio.Channel#select method, which according to documentation, blocks forever when passed 0.
|
done |
|
Still some rubocop issues, can you please fix :) Thanks. |
|
btw, the spec that I added blocks indefinitely, that is, without the fix inserted in the subsequent commit. The around wrapper being used in celluloid specs here would have prevented this and provided better visibility. @tarcieri not saying that all of those wrappers should be imported here, but some of them would definitely make sense, specially for travis PR error visibility. |
|
Released as 1.2.1 |
Solves hanging actors problem described in socketry/timers#58