Slimelite BBS

Re: orig MSGID -> new REPLY

Back to FREE25

From HECTOR SANTOS · 1996-04-12 18:15 · #73179 in FREE25

In reply to Re: orig MSGID -> new REPLY by imb@asstdc.scgt.oz.au

HS> Second, lets consider that the INTERNET is a somewhat of a controlled
HS> distribution system and dramatically differs from the fidonet
HS> distribution : structure.

II> What prevents the propagtion of duplicates is the fact that NNTP
II> presents the message-id first and waits for either a "send-me"
II> (indicating that it hasn't seen the message before) or a refusal
II> response (to say that it has).

I'll say it again, it is "somewhat of a controlled distribution" and
Fidonet doesn't have the priverledge of single source verification of an
identification.

Once everyone moves Fidonet via the internet, then we will be able to
get access to daemons that allow us to check for duplications.

II> People still quite happily use Windoze even though the DOS command-line
II> offers significantly superior performance. They do this only because
II> they believe there is some functional improvement (although I can't
II> think why :-)).

If you can't think why, than maybe you shouldn't "presume" so lightly :-)
I understand your point though.

II> However, the resultant product generally has a better than average
II> maintainability and far fewer "undocumented features".

You sound like we all once were - "idealist" <g>.  During my early
engineering years, my boss had a saying for me "What something done
tomorrow, give it to bob. What something done right, give it to Hector".
:-)

I can argue against your statement, but I would only say that
practicality and maintainability are not necessarily inversibly
proportional to each other as you make it out to sound.

Your statements are made from an Internet perspective, and FidoNet is
not the internet.  We can learn ideas from the internet, and although I
tend to believe the Internet is the one learning from everyone else, the
bottom line is that the two email systems is not the same.

II> However, I'm not saying that FidoNet should suddenly switch tomorrow. I
II> think everyone accepts that this is impossible. What should happen is
II> that all new software be written to implement FTS-0009.

But you won't be able to mandate it. And even if it were to change, I
can assure you it will take on an WHOLE different meaning and address
more than just 'serialization', but for addressing the participants of a
message.

II> In fact, this was the intent of its adoption as an FTS .. authors are
II> expected to "improve" over time :-)

or make due with what there with a cavert here and there. :-)

HS> Where would you get the DOMAIN from the message?  Must FTS-9
HS> change the address have the domain in it?

II> This is the next piece of FTS-0009 that needs clarification.

Buts its not in there, hence it will not work.

II> You simply don't get the instance of a node being known as
II> 99:12/34@dom1 and 99:12/34@dom2 .. it just doesn't happen.

Well, technically you don't need an specific domain. You can use the
ZONE for a fido address and have a ZONE vs DOMAIN look up table.
However, the problem here is that ZONES are not unique to domains
and FTS-9 does not state what exactly is a "Address".

II> :    You have DATABASE lookup #1 to find a matching ECHO TAG.

II> The intention, in my example, was that echos are readily looked up
II> using a straight index into a table. You can optimise this and the
II> comparison of the domain by comparing all the other data elements
II> first and, only when a true match is found, comparing the things that
II> are expensive to look up. That's why I list them at the end of the
II> structure.

But you need to look at the issue of huge tossing duties, of the likes
of 10,000 messages per day as many Planet Connect hubs do.  Oh, you're
in australia. So you don't see this happening.  I just talked to Boyd
Goodin at Planet Connect, he says PC will be available in Europe anytime
now. I'm sure it will come to down under eventually.

In fact, Boyd's dupe checking is practically turned off!

HS> 4) What is the table capacity on a TOTAL BASIS and on PER ECHO BASIC?

HS>  The 3 year factor of the FTS-9 specification now comes into place
HS>  now.

II> Oh no .. duplicate databases on the internet are rarely held for
II> longer than 14 or 30 days ..

Didn't know that... You see, Internet people are reasonable too. :-)

II> Likewise, we should only
II> expect to maintain duplicate information for reasonable periods.

Exactly, which comes down to my CRC solution which is "reasonable".

II> If a "reasonable" propagation time for a message is, say 14 days, then
II> we simply keep 45 days worth of dupe history and refuse to propagate
II> messages older than that. This last bit is important !

Very important.

II> Do you really want to keep information about messages beyond their
II> use-by date ? Most have ceased to be relevant within 7 days.

Thats not the point.  As Moses said "So it is written...So it is done"
or something like that... :-)   People here have argued the facts of
what is a serial number or what is a address and how to use FTS-9 under
a "Follow up" and they all pointed to me saying "it is the LAW" that is
how it is written.  My point in this whole discussion is that the DUPE
checking and threading aspect is not "reliable" for the many reasons we
discussed and the ONLY real reliable information is the addressing
portion of the specification.

If you look read the message 2 years later or come across it, saved away
in your FOLDER, the ADDDRESS is the only thing that is "useful". Surely
not the serial number.

That is my whole point with FTS-9 and why I say, that if we are going to
change it, we BETTER make sure the ADDRESS is made a significant
redesign consideration.

II> Another point .. what happens when someone manages to clobber their
II> sequence number container .. do we now refuse all his/her messages
II> until they post one that exceeds our previously recorded "high-water
II> mark". The "test of reasonableness" says otherwise.

Well, thats where I said in one of my message bases, what happens if
the SYSOP deletes his DUPE file... it happens ALL the time in the
HOBBY world.  Can you imagine telling a hobby sysop "Thou shall never
delete your DUPE file, if you do, you will voilate FTS-9!" <grin>

II> .. but this results in false collisions as discussed if you use the
II> CRC as the only mechanism of comparison.

But thats not the case, and I would venture to say that even with a
first level CRC check, it is extremely "reasonable" as a search into a
"bucket". The odds are very low of erroneous dupes and practical
experience has proven this.

I would love to finally see two pairs of following data in my limited
DOMAIN of messages on my local machine:

          TOWHOM      +
          FROMWHOM    +
          SUBJECT     +
          ASCIIZ DATE +
          ECHO TAG

that yield the same CRC.  I just want to see it :-) I have not seen
it and when we do, it will be rare and remember, thats assuming the
other comparision elements were passed.

II> You'd be surprised at what they're teaching "kids" these days .. <bg>

Well, I do know one thing, it is highly competitive out there and
overcrowded.  Those where practical experiences will have an advantage.

BTW, my next career move is computer law <g>
--- Platinum Xpress/386/Wildcat! v1.2j
 * Origin: Xpress Tech Support BBS * 305-248-7815 (1:135/382)