Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] Making an embeddable libev
From: Marc Lehmann (blacklist_at_[hidden])
Date: 2007-12-19 20:27:32

[cc sent to openmpi: please, openmpi developers, I really like your
project, but if that guy is part of your development effort, please send
somebody else who is not trolling on our mailinglist. I am very open to
working together with you in ironing out any issues you might encounter,
but spreading FUD about libev openly is not something I look forward to in
the future. Thanks for your attention].

[courtesy cc of replies appreciated, I am not on the list]

On Wed, Dec 19, 2007 at 07:38:27PM -0500, Jeff Squyres <jsquyres_at_[hidden]> wrote:
> I am saddened that you have responded with such ferocity -- calling me
> a liar and making several other derogatory statements.

You itself admitted to acting like a dork and an idiot, and unfortunately
you didn't stop with your ignorance policy. It is nice and remarkable that
you apologised, but that doesn't give you a license to continue trolling.

If you are saddened, maybe there is a learn-effect to act less like an
idiot in the future, but somehow I doubt it from reading this mail.

You could have easily avoided the ferocity by not using your rethorics
(i.e. not being a liar), reading the documentation and listening to the
expert willingly explaining all of it.

The fact that I wasted any time on you by patiently explaining thigs to
you again and again makes me not sad, but rather annoyed. People like you
are a bad thing for software projects, and I can only hope you either act
better when working on openmpi or that openmpi finds a way to route around
your damaging behaviour.

> Your docs state:

   "The goal is to enable you to just copy the neecssary files into your
   source directory without having to change even a single line in them,
   so you can easily upgrade by simply copying (or having a checked-
   out copy of libev somewhere in your source tree)."

Indeed. And so far, the goal has been met **overwhelmingly well**, **to
the letter of the statement** (my emphasis), as *proven* by a number
of projects who do *exactly that*. Why you keep spreading fud about it
escapes me, but its an extremely unfair and hostile behaviour by you, and
something I will certainly not tolerate.

The reason you keep repeating the opposite or cast doubt on the validity
of that statement is the reason for the perceived "ferocity": you cannot
expect that repeating unsound arguments (or worse, spreading FUD) will
somehow make somebody else give in (except maybe to stop it by ignoring an

With me, it only has the effect of repeatedly explaining you why it's
wrong, each timer longer and with more force, as for some (never stated)
reason you never accept those arguments, nor did you ever explain your
reasons (or reasoning, if there is any) on why. To me, it decidedly looks
as if you only tried to push something into libev that you would like to
see, just because you like to see it, wether it makes technical sense or
not. You actually stated that a lot in this thread.

If you find this saddening, maybe you should start listening to what
people tell you, instead of dismissing it and dumbly repeating your claims
without delivering any evidence.

But if you keep spreading FUD about libev then you must not be surprised
if somebody else reacts accordingly by calling you a liar, because thats
what you are, after all.

> Having a checked out copy of libev (or, more precisely, an expanded
> tarball) is exactly what I want.

It works for lots of other projects, and it would work for openmpi, as has
been explained many times.

> And I wanted to use your

And I explained why this is not a good idea (mostly because it doesn't
work). The fact that you ignored all that and keep repeating that you wnat
to do the wrong thing despite being told repeatedly that it is not the
right approach

> I have tried to explain why, I have given code examples,

Thats an outright lie, I asked you many times to explain why, but you
never explained it. You repeatedly said "you want this" and "you need
this", but these are not reasons, which you did not deliver. The exampels
you gave were either incorrect, didn't solve the problem as claimed, or
degraded the quality of the library (by complicating and fixing things
that are not broken) or some combination thereof.

And as has been explained a few times, you can easily embed and do
everything you want with libev as is, as a lot of work has been invested
to make it easy. The fact that you disregard this work and make factually
wrong claims about it will also not have any positive effetc on your
credibility, of course.

> and I have cited that it's a common AC/AM/LT model.

Bullshit, embedding other libraries that way is the absolute exception. I
know of no such library, and I know a lot of libraries. Calling it
"common" is outright wrong.

> I've even been polite in the presence of overwhelming hostility.

Thats also not true. Not answerign questions, not reading the documentation
and making false statements publicly is not polite, it is, in fact, very
hostile of you and has been noted as such.

> I freely admitted that I got my first posts all wrong and I totally
> bozoed by missing the docs. But somehow, my asking for a feature has
> resulted in rather fierce attacks.

Thats bullshit, you did not ask for a feature, but you demanded a change
that would not be a feature, but a bug, as has been explained. Asking for
the reasons never resulted in anything but claims, never any reasons.

> I have come to the sad conclusion that your hostility makes further
> discussion pointless.

You are insulting me here for no reason. I am not hostile at all. If anybody
is hostile it is you because you do not want to argue logically or listen to

Your behaviour might indeed make further discussion pointless.

> FWIW, thanks for the software. It will probably help us quite a bit.

Good to hear that. If there are problems, feel free to report them. Best send
somebody else who a) has a basic level of understanding of unix and b)
doesn't insist on changes just because he wants them without reasoning why.
You will find that very few maintainers that want to kepe a certain quality
level in their released packages will give in to gratitious change requests.

> On Dec 18, 2007, at 8:59 PM, Marc Lehmann wrote:

And top-posting is also not a sign of expertise, although some people do it

                The choice of a       Deliantra, the free code+content MORPG
      -----==-     _GNU_    
      ----==-- _       generation
      ---==---(_)__  __ ____  __      Marc Lehmann
      --==---/ / _ \/ // /\ \/ /      pcg_at_[hidden]
      -=====/_/_//_/\_,_/ /_/\_\