are new benchmarks needed?
schmorp at schmorp.de
Mon Jan 10 08:22:51 CET 2011
I abstained from this thread until I found time to read it and make a
preliminary benchmark. I will not make further comments on this thread
- anybody has anything to tell me about this topic, please start a new
thread, I have better ways to waste my time.
On Tue, Dec 28, 2010 at 10:40:58PM -0600, Charles Kerr <charles at transmissionbt.com> wrote:
> (1) When libev compares itself to libevent, it should use reasonably
> up-to-date information. If libevent2 is better than libevent1, as it
libevent has always been better than libev in those areas where it is
better. thats rather logical, no?
obvious things where libevent is better out of the box include dns and
http. also, these stream buffers are really nice to have. etc. etc.
Maybe to you "better" equals "speed", in which case I can sell you a
few faulty but fucking fast programs. I doubt that though, I think you
equate "better" with something that is not speed, in a thread about
benchmarks. Whats your point in introducing fuzzy concepts such as
I always encouraged people to make up their own mind, and I will continue
to do so. You will not get a statement out of me where I unconditionally
say "libevent is better than libev" (or it's opposite), because it would
be dishonest, and a disservice to my users.
> seems to be, then continuing to omit libevent2 /is/ misleading,
> whether intentional or not.
What a load of bull. The benchmark compares against an old version of libev
as well, there is no "continuing to omit libevent2" because the benchmark
wasn't updated to newer libev versions either. It's also not "continuing"
because libevent2 had only been released a few days before your mail.
For some reason, I am sure that you will not complain that the new
benchmark graphs will continue to omit newer libev versions. Nor did you
complain about just that in the past.
Implying that anybody would "continue to omit libevent2" is just trolling,
and sorry, but I feel that is what you do with your implications and
Again, even though that has been explained to you before, the benchmark
document explains clearly what versions of libevent (and libev) it refers
to, and carries a prominent date tag at the top.
And yes, pressuring me for benchmarks by implying that I would publish
misleading statements (whether intentional or not) about libevent2 is
fundamentally dishonest. It might be surprising to you, but I do have
better things to do than to benchmark for you as soon as a new version of
libevent2 comes out.
Especially, as can be seen from the preliminary results, nothing really
drastic has changed, so the time to prepare bechmarks was essentially
wasted, and rewriting (and checking!) the document would be essentially a
waste of time.
> (2) If there are no practical performance difference between libev and
> libevent, as libev's author says,
If I said that you can surely point out where I said that, because I am
not aware that I did.
Needless to say, I am a bit pissed about your repeated misstatement of
my intentions or of what I said. Please don't put words into the mouths
of other people that they didn't say, especially not when they twist the
meaning to something rather different.
> I find that incomprehensible
And I find it hard to understand why you bother to make so much trouble
for nothing. The way you do it, by implicating me to have said or done
things I didn't do, is simply inacceptable.
The choice of a Deliantra, the free code+content MORPG
-----==- _GNU_ http://www.deliantra.net
----==-- _ generation
---==---(_)__ __ ____ __ Marc Lehmann
--==---/ / _ \/ // /\ \/ / schmorp at schmorp.de
More information about the libev