Hello libev team,
I realize that doing multiple ev_async_send's on a ev_async watcher
(before the previous ev_async_send is processed) causes the callback to
be invoked only once by design. I would like for a separate callback to
be invoked each time an ev_async_send is called on the watcher. I
On Tue, Oct 23, 2012 at 01:16:11PM +0530, "Chaliha, Sayan"
wrote:
> be invoked only once by design. I would like for a separate callback to
> be invoked each time an ev_async_send is called on the watcher. If this
> is not a feature you would put in production, I would be very grateful
> if you c
On Wed, Oct 24, 2012 at 04:20:48PM +0200, Marc Lehmann
wrote:
> On Tue, Oct 23, 2012 at 01:16:11PM +0530, "Chaliha, Sayan"
> wrote:
> > be invoked only once by design. I would like for a separate callback to
> > be invoked each time an ev_async_send is called on the watcher. If this
> > is not
Hi,
I'm heavily abusing CLOEXEC on a lot of file descriptors to fork &
execve into a sandboxed mode for handling connections (for security
reasons, as additional exploit mitigation of sorts).
Furthermore I'm using vfork for performance reasons and therefore cannot
close the event loop by legitima
On Wed, Oct 24, 2012 at 04:25:32PM +0200, Georg Wicherski wrote:
> - Can you set CLOEXEC on the epoll FD, please?
libev has always set CLOEXEC on the fd, AFAICR - which version are you
using?
> - What is a hackish way to get that FD and do this myself until such a
> patch is available in the maj
On Tue, Oct 23, 2012 at 04:47:02PM +0200, Meir Yanovich
wrote:
> is there some kind of advance examples or tutorials for using livev as tcp
> server ?
Well, the best "advance(d) examples" is usually production software.
rxvt-unicode shows how to embed it, and also the C++ API, and there is a
bun
Georg,
you can give a shot for valgrind --track-fds=yes.
It might give you some clue about where was fd created.
May be it's not libev specific.
2012/10/24 Marc Lehmann
> On Wed, Oct 24, 2012 at 04:25:32PM +0200, Georg Wicherski
> wrote:
> > - Can you set CLOEXEC on the epoll FD, please?
>
>