abnormal termination fails silently + burns cpu

[saluzafa]

fix: use @pipe.read_nonblock(1) instead of @pipe.read(1)
fix: rescue EOFError (raised by #read_nonblock) and wait 1 seconds before iterating over

[thibaudgg]

Hi @saluzafa, thanks for your pull request, this is much appreciated.

Could you please add some more details are both fixes explaining why they needed and which issue they are particularly addressing.

If not too complicated, adding specs would be also great.

[sbleon]

+1 for merging this change. It prevents 100+ CPU utilization by fsevent_watch when I start my Rails 5 project with bin/rails s. I'll be using @saluzafa's fork for now.

[thibaudgg]

@sbleon that's great to hear, I'm also looking merging that PR!
I would just like to see a bit more context in the commit message.

[saluzafa]

Hello there!
When a process running fsevent_watch is abnormally terminated, @pipe.read(1) hangs and try to read data from a pipe that is actually closed. To prevent this behaviour, I tought of using a non-block read. When #read_nonblock can't read anything it raises an EOFError exception, so I'm catching this exception and the process terminates properly.

I'm going to remove the useless part rescue EOFError [...] since EOFError is a child of IOError (handled at line 82).

Have a good week-end :)

[thibaudgg]

@saluzafa great, thanks for the explanation!

[saluzafa]

Welcome :), I just noticed that I forgot to bump the version number, are you going to do it by yourself later or do you want me to do it now? Cheers!

[thibaudgg]

I just did it, version 0.10.3 is out!

[sbleon]

Thanks so much, y'all!

…

On Sat, Mar 3, 2018 at 2:29 PM, Thibaud Guillaume-Gentil < ***@***.***> wrote: I just did it, version 0.10.3 is out! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#81 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAMIN5OJ6I8EuCLOj4FLrUlJApUcTE7Oks5tau8fgaJpZM4SWuyH> .

[padi]

I've held on commenting until I've made installed 10.3 and observed for a few weeks.

When I use spring (which this gem uses) in a Rails 5.1 project, I still get 100% cpu usage some of the time.

Not sure how I can falsify other than occassionally seeing 100% cpu in ruby. Do I need to update to the latest macOS? I'm currently in macos 10.12.6

Thanks so much for possibly resolving this long-time issue!