[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [Libevent-users] Asynchronous writes in the event loop model



On 1 feb 2012, at 23:31, Nick Mathewson wrote:
> … [ questions about POSIX signals and pthreads ] …
> …
> Sadly, this whole business is a huge pile of gunk.

Thanks for the epic reply, Nick.

I did some additional research before diving into my POSIX signal plan and it indeed looks like a huge mess. My initial optimism about potentially having my pthread-cake and eating it too waned quickly. Especially on Windows you enter some bizarrely limited alternate reality regarding signal handling.

For now, I've settled on using a loopback socket for interthread communication. I've not yet tested this approach with thousands of clients, but I will certainly benchmark later comparing sockets to pthread mutex locking. You are probably right that the pthread mutex scenario wouldn't be too bad after all.

Once I get some concrete numbers to share, I'll post them to the list.

With kind regards,

Frank

***********************************************************************
To unsubscribe, send an e-mail to majordomo@xxxxxxxxxxxxx with
unsubscribe libevent-users    in the body.