• Nodelist global

    From Felipe T. Dorado@2:345/702 to Todos on Mon Aug 13 11:42:50 2001
    Hoal Todos :)

    Para aquellos interesados en Nodelists, especialemente la global.

    * Forwarded by Felipe T. Dorado (2:345/702)
    * Area : fidonews (fidonews)
    * De : FidoNews Robot, 2:2/2 (Lunes 13 Agosto 2007 01:41)
    * a : All
    * Tema : FidoNews 24:33 [02/05]: General Articles =============================================================================

    Interzonal nodelist discrepancies
    By Michiel van der Vlist 2:280/5555

    In an article in FidoNews #24 of this year, ex-ZC6, Carol
    Shenkenberger announced the final demise of Z6 per july 26th. On that
    date the last entry in Z6 was removed and a placekeeper was
    substituted for the zone segment. No, this article is not about the
    usefulness of the placekeeper. It is about another observation that it triggered. July 26th is a Thursday, so if all went well, the change
    should make in in the nodelist of next Friday, #208.

    Well, it didn't. Not in the version of the nodelist that is
    distributed in zone 2 anyway. So I got myself a copy of the Z1 and Z3
    nodelists #208. And yes, the placekeeper was there. The change did not
    appear in "my" nodelist until #215. So the Z2 nodelist is one week
    behind that of Z1 and Z3? For this change yes. But there seems to be
    more to it than just a simple one way delay.

    On Monday July 30th I submitted a change in the net 2:280 segment. Two
    nodes, 1026 and 4312 that had been flagged as Down for some time, were
    removed. The change made it into "my" nodelist #215 next Friday. But
    not into the nodelists as distributed in Z1 and Z3. It did not make it
    into Z1 until a week later: #222. So it seems the "own zone" is always
    a week ahead?

    No, because an even more intriguing discrepancy is that 1:275/311 is
    listed as Down in Z2 nodelist #222. But not in any of the Z1 versions
    that I laid hand on. This can not be explained by a mere delay. The
    only explanation that comes to mind that a change was submitted and
    later retracted. The change was processed, both for the Z1 and the Z2
    list. The retraction however only made it in time to annul the change
    made into the Z1 nodelist, but not in time for the Z2 nodelist.

    Someone unfamiliar with the internal workings of FidoNet might easily
    conclude that it is the International Coordinator that issues the
    FidoNet nodelist and that therefore every member of FidoNet has the
    same nodelist. Well, it isn't so. It may have been done like that in
    the very beginning, but soon - if not right away - after zones were
    introduced the ZCs started issuing their own versions of the nodelist.
    They gathered the partial zone segments from each other and assembled
    their own version of the nodelist from them. This made sense in the
    days when sending files to another continent was expensive and slower
    channels than a direct ZC to ZC contact were used to reduce the cost.

    The versions distributed in the different zones are not the same. The
    most obvious difference is that the zones lists their own zone first.
    This was introduced by the first ZC2, Henk Wevers. His own mailer
    Dutchie did not use an index file to access the nodelist, it did a
    simple linear search. Since most lookups were within the own zone,
    putting that first speeded up the process of nodelist lookup. Other
    zones followed this practise.

    All this made sense at the time. But now, when we have almost
    instantaneous global communication at next to zero cost? A change I
    submit on Monday, does not make it into the other zone's nodelists
    until Friday next week, ELEVEN days later?! In 2007? Oh c'mon guys, we
    can do better than that!

    Why do we still have a different nodelist for each zone? All the
    reasons for doing it in the past no longer apply. Why not let the
    International Coordinator assemble a list and distribute that as *the*
    FidoNet nodelist. The only obstacle I see is that we do not seem to
    have an IC at the moment. There isn't a zone/1000 entry - the
    traditional AKA if the IC - in any of the five zones of FidoNet that
    remain. None of the ZC's seem to claim the position.

    Still, even lacking an IC, there is no excuse for all these delays.
    There is no technical reason that I can see, why a change on the net
    level that is submitted on Thursday afternoon by an NC, can not be
    included in in the nodelist that is available to every Fidonet member
    on Friday morning. What it takes is a bit of interzonal coordination.
    Something that never was the strong point of FidoNet, to use an
    understatement. So guys at the top of the *C hierarchy get your act
    together and do what it says on your hat: Coordinate!

    (c) 2007, Michiel van der Vlist. All rights reserved.


    -+- Azure/NewsPrep 3.0
    + Origin: Home of the Fidonews (2:2/2.0) =============================================================================

    Felipe :)

    --- GoldED+/LNX 1.1.5
    * Origin: El Zoco BBS, COrdoba, Espa-na, UE, Tierra, Sol, VL. (2:3 (2:345/702)