are new benchmarks needed?

Pedro Melo melo at
Wed Dec 29 08:00:02 CET 2010


On Wed, Dec 29, 2010 at 4:40 AM, Charles Kerr
<charles at> wrote:
> (1) When libev compares itself to libevent, it should use reasonably
> up-to-date information. If libevent2 is better than libevent1, as it
> seems to be, then continuing to omit libevent2 /is/ misleading,
> whether intentional or not.

I'm sorry, but while I can see your point that libevent2 is much
better than libevent and therefore could be on par with libev, and
that marketing and real quotes sometimes don't add up, you are wrong
about the benchmark page: they clearly states that this is a
comparison between libev and libevent, not libevent2.

You, as a software developer, would prefer a comparison between the
latest versions of each, and I agree that it would be nice to have,
but neither is that a requirement for the project, like something that
it must do, nor is the current benchmark wrong.

You seem to be evaluating both to make a decision for your software
project. I think you are in a perfect condition to write a article
describing your own comparison between libev and libevent2. (May I
suggest that you use the same layout as the current libev benchmark?).
After you publish your own results, its up to Marc to link (or not) to
your benchmark from the libev page. If doesn't agree with your
methodology or with your results, its up to him to do the work

Pedro Melo
xmpp:melo at
mailto:melo at

More information about the libev mailing list