Understanding Systemd's Event Loop API

[phoronix]

Phoronix: Understanding Systemd's Event Loop API

Lennart Poettering has put out a new blog post today to better explain sd-event, the systemd Event Loop API...

http://www.phoronix.com/scan.php?page=news_item&px=Systemd-Event-Loop-API

[unixfan2001]

Wohoo!

Any minute now I'll be able to browse through a bunch of new "Lennart Poettering is the devil incarnate" hate tirades.

[wodencafe]

Wohoo!

Any minute now I'll be able to browse through a bunch of new "Lennart Poettering is the devil incarnate" hate tirades.

White knight. :P

[gens]

how about implementing a run-of-the-mill event system using clock_gettime and usleep instead of faffing around with fds ?
idk if that is what they changed now, the blog post doesn't say
(can be combined with poll-s for external events, ofc)

[clockley1]

I look forward to using sd_event and sd_dbus in my programs.

libsystemd replaces libevent, libdaemon and libdbus. The consistent API makes it easier for developers, and fewer dependencies means less time typing yum install/apt-get to install packages from the source code.

[Hibiki Kanzaki]

I look forward to using sd_event and sd_dbus in my programs.

libsystemd replaces libevent, libdaemon and libdbus. The consistent API makes it easier for developers, and fewer dependencies means less time typing yum install/apt-get to install packages from the source code.

Me too, I'm looking forward to eventually just having a liblennart as the only only dev package and library I have to install and use. In time hopefully it will also replace libc so there's only one library.

[pal666]

how about implementing a run-of-the-mill event system using clock_gettime and usleep instead of faffing around with fds ?
idk if that is what they changed now, the blog post doesn't say
(can be combined with poll-s for external events, ofc)

events are all about fds
return to your mom

[liam]

And yet another reason to attempt to implement iocp in Linux.
Yes, it's very hard to do, but it's the CORRECT solution, and fixes other, STILL outstanding problems with event monitoring.

[computerquip]

And yet another reason to attempt to implement iocp in Linux.
Yes, it's very hard to do, but it's the CORRECT solution, and fixes other, STILL outstanding problems with event monitoring.

I agree. Boost Asio wraps over epoll but its dirty and showcases some issues.

[skierpage]

"When creating timer events a required accuracy parameter may be specified which allows coalescing of timer events to minimize power consumption. For each clock only a single timer file descriptor is kept, and all timer events are multiplexed with a priority queue."


Good stuff. So is The Devil Incarnate being forthright about the limitations of, and alternatives to, his offspring. He does such a good job acting as a thoughtful productive Linux developer, that only the knee-jerk haters are able to see through it.