HOW does RSS work?

HOW does RSS work?

Post by » Mon, 09 May 2011 20:51:28

Don't you get pissed-of when these on-line explanations
just talk about "It's a better EXPERIENCE and 'life is better
with coke'..".

I want to speak with the injun-year not the sales-clerk!!

The folowing level of detailed explanation for USEnet is what I need,
to evaluate RSS:
* the messages are regularly distributed to the servers, which
  have a mechanism to update only messages which are new:
  each message has an 'IDnumber'.
* the clients post their messages, specifying the Group/s and
* the clients can d/l N of the newest message-headers from groups
   which are subscribed to.
* after the client-user has read and selected any header-bodies,
 which seem potentially interesting, these selected messages
 [total of head and body] can be d/l-ed; possibly at a later
 on-line session.
* since each message has an IDnumber and the IDnumber of the
message that it is replying to, if it IS a reply, the messages of
the original 'Subject' can form a tree; since multiple replies
to a single message, correspond to the branching of the node
into descendants.

In contrast, blogs offer NO indexing or selection within a
single thread: the whole, perhaps 200 messages must be
d/l-ed TWICE if you want to ask a question and check for a
reply.  OTOH USEnet requires only:
d/l-ing the header tree and identifying messages that are
answers to YOUR query, within the message tree.

The result is that blog-commenters of popular blogs,
post their comments 'early' without having time to
sleep over ideas, and consequently give only superficial
input. I.e. the early commenters pay' for small traffic-loads,
and the last commenter has to pay 'the biggest d/l.

So what are the corresponding mechanisms of RSS?



HOW does RSS work?

Post by Tobias Blas » Mon, 09 May 2011 22:12:32

["Followup-To:" header set to comp.os.linux.networking.]

Quote:> [snip]
> So what are the corresponding mechanisms of RSS?

You can find the complete RSS specification here:


1. NIS fallback to DNS on RS/6000 AIX 3.2.x; can it be done?

|> >  
|> > >Is there a way to have NIS also search the institutional DNS
|> > >database for host names?
|> >  
|> >     Yes, on the server side only.  If the server /etc/yp/Makefile is
|> >     changed to use 'makedbm -b' for the hosts.byname and hosts.byaddr
|> >     maps, (the -b is the new part) then ypserv will ask DNS if it can't
|> >     find the answer in it's own maps.  As far as the client is
|> >     concerned, it appears to be still always talking to the NIS server.
|> But what about name look ups on the server? There, a /etc/resolv.conf
|> needs to be present, which means name look ups ignore NIS, go first to
|> DNS, then to the local /etc/hosts file. Clients look first into the yp
|> hosts map, then ask DNS, which means the order is reversed.
|> And what about slave servers? Do they need a /etc/resolv.conf? Wouldn't
|> they then need a local /etc/hosts file too? That seems to defeat the
|> purpose of NIS somehow.
|> Please tell me I'm wrong.
|> > - --
|> > - --

My experience is that you are not wrong.  I tried the suggested change to the
make file and found that I also had to remake YP on the slave servers.  So,
the slaves wouldn't get names from the Master NIS server but from the
DNS server that is not under my control.  This also made a mess when it came
time to rsh into different systems because on the clients .rhosts could use
just the name but on the slaves you had to use the name + the domain.  This
really messed up a lot of rcp etc. stuff that was being done on the net.  So
I had to go back to just NIS and eliminate DNS from my net for now.

I'm thinking of trying various domain names.  It is my understanding that
you can have the NIS domain be one additional part of the DNS domain.  (ie
DNS a.b.c and NIS nis.a.b.c).  I'm not sure how this works though.  Another
option is to set up DNS on the NIS master server.  This way the slaves would
get the correct name from the same host, although that host will have two
different files with the host names in it.  So I thought of looking at making
the NIS master server a DNS cache-only server.  I don't know what that will
accomplish though.  

As far as the order in which the clients look for names, I once posted some
direct quotes out of the system manual that said that it did it one way and
then ON THE SAME PAGE said it did it the other.  

This shouldn't be this hard, but I admit to being totally confused by it.  Since
no one provided me with an answer when I was asking these same questions a
couple months ago, I have to assume that no one else has solved the problem

Then again, maybe its a FAQ and I just haven't found it yet.

2. Print sharing with Mac

3. DOS on the RS/6000

4. get unread mail only

5. Maxtor drive on an rs/6k: anybody done it?

6. (no subject)

7. NIS fallback to DNS on RS/6000 AIX 3.2.x; can it be done?

8. XF86Config for Compaq 5380 laptop

9. How to make pcmcia ethernet card that works with Linux work with dos? . .

10. HELP: Connectivity between DOS/DOS and DOS/Linux

11. Re. SPARC to RS/6k Re. SPARC to RS-6000

12. Sorry!! RS/6000 instead RS/2000 ;-)

13. set RS-423 to RS-232