"DBus Embedded" - a clean break

Havoc Pennington hp at pobox.com
Fri Jan 21 07:12:30 PST 2011


On Fri, Jan 21, 2011 at 8:16 AM, Simon McVittie
<simon.mcvittie at collabora.co.uk> wrote:
> I did look at using libev, but I was saddened to find that it uses
> floating-point time values, which isn't really going to fly on the platform
> I'm doing this for (Maemo/MeeGo on ARM).
> It also seems rather perverse to go from int milliseconds (libdbus) to
> floating-point (libev) back to int milliseconds (poll/epoll/select)...

Weird. I'm surprised this matters enough to overcome the overall
advantages of libev. (It is FAR faster than glib's main loop for
example in my benchmarking, especially at adding/removing sources, and
timeouts in particular scale better. Granted, with a small number of
sources it hardly matters.)

> The fact that its error handling is basically abort() also puts me off.

Surely that's just for programming bugs.

> It looks as though the main loop refactoring I'm doing at the moment would
> be necessary for either epoll or libev, FWIW.



More information about the dbus mailing list