Newsgroups: comp.parallel
From: eugene@nas.nasa.gov (Eugene N. Miya)
Subject: Re: Why things don't post for a while
Organization: NAS - NASA Ames Research Center, Moffett Field, CA
Date: Fri, 6 Jan 1995 18:18:30 GMT
Message-ID: <D1wzJs.C1w@cnn.nas.nasa.gov>

In article <D1ABH5.231@cs.dal.ca> Erik Demaine <edemaine@ug.cs.dal.ca> writes:
>The reason for this is the moderator.  The moderator... someone at Acadia...
>goes through all posted articles before they show up on the newsgroup.  As
>far as I can see, *everything* is posted.  He makes comments when he can to
>answer some questions quickly (they look like [Mod note: ...]).
>
>Sometimes it takes a couple days, but usually things are quite fast.
>
>[Mod note: I try to to the posting at 4:30pm everyday. There are a few days
>where I have to work long, and might pass on posting for a day. When I receive
>identical postings separated by a few days, it's difficult to remember that
>there was already one posted. For postings that are duplicated and are received
>by 4:30 on the same 24-hour period, I usually do remove one from the group.
>-richard]

I would suggest not being so quick to note the moderator.  Reliability
in parallel/distributed systems is one big issue which is very misunderstood,
and the usenet is such a distributed system.  For instance, people
interested in big M/G/T/FLOPS tend to neglect fault tolerance and reliability
and many people make the mistake of assuming that more hardware means
more reliability.  This is why so many parallel systems fail.

Posts have to first make it to whosoever is moderating, and then after
approval must then make it out.  And of late network propagation is
getting worse.  Ways exist to measure it in this group (auotmated)
but they are not in place.  No one should assume the net is reliable.

I have forwarded posts and never seen them on this machine, but got ACKs
from the moderator or saw them on other machines.

Don Norman in a book I am currentl reading notes to avoid systems which are
quick to fault the user or people involved.

As Crispin Cowan notes in his sig:
"A distributed system is one in which I cannot get something done
because a machine I've never heard of is down"   --Leslie Lamport


--eugene miya, NASA Ames Research Center, eugene@orville.nas.nasa.gov
  Associate Editor, Software and Publication Reviews
  Scientific Programming
  {uunet,mailrus,other gateways}!ames!eugene
Seeking Books to buy:	Bongard, Pattern Recognition
			3 down 1 to go.


