chicken-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Chicken-users] Chicken Gazette - Issue 10


From: Alan Post
Subject: Re: [Chicken-users] Chicken Gazette - Issue 10
Date: Tue, 2 Nov 2010 09:53:33 -0600

On Tue, Nov 02, 2010 at 03:25:50PM +0100, Jörg F. Wittenberger wrote:
> Am Dienstag, den 02.11.2010, 14:52 +0100 schrieb Moritz Heidkamp:
> > Hi Jörg,
> > 
> > Jörg "F. Wittenberger" <address@hidden> writes:
> > > Though one thing I haven't been able to figure out: is there a way
> > > encrypt and sign zmg messages?  Or would I have to add that on top?
> > 
> > funny you ask, this has also come up on the ZMQ mailing list
> > recently. ZMQ is just a transport/messaging system, it doesn't try to do
> > more than just that and guarantee you that only complete messages will
> > arrive at your end. What you put into the messages is up to you.
> 
> That's tough, unfortunately.
> 
> In the BALL/Askemos code base, the source signature (x509 only by now)
> of is an important property used to gain confidence about the current
> network wide state.
> 
> > I have an extension for the mailbox-threads egg which adds support for
> > remote messaging between thread mailboxes via ZMQ. I'll check it in
> > later today, maybe it is useful for you!
> 
> Given no trust in the message source, I'll only be able to replace some
> low level code.   Is it true (that's how I read the egg API) that ZMO
> will only distribute/collect messages?  By "only" I mean "not do any
> removal of re-transmits, no counting of majority of messages delivered
> or any other handling of synchrony issues"?  (I.e., is it really just a
> transport level thing?)
> 
> From skimming across the zmq website one more question.  I always asked
> myself what the point of the poll(2) call is.  More efficient that
> select(2)?  Apparently that's the case.  ...  Which, given that I have
> some Scheduler patches still pending, I'm almost about to switch those
> to use pool instead of select, to make both a one time effort to
> integrate.
> 

My understanding is that unpacking and repacking FD_SET objects in the
select call is the reason it is slower than poll--that the interface
is more poorly designed.

-Alan
-- 
.i ko djuno fi le do sevzi



reply via email to

[Prev in Thread] Current Thread [Next in Thread]