From Leonard Erickson · 1996-04-09 11:16 · #73028 in FREE25
In reply to orig MSGID -> new REPLY by Pierre Phaneuf
-=> Quoting Pierre Phaneuf to imb <=-
PP> As a tough, what's the exact algorith for Internet serial numbering?
There *isn't* one. There's just the requirement that MEssage-ID be
unique, and a restriction on the allowable characters. No spaces, no
angle brackets, etc because those mean other things when parsing header
lines.
Here are the Message-ID lines from the messages I have in my
news.answers directory. Some are rather non-standard, since they are
all FAQs (unfortunately, I just purged the rest of the stuff after
reading it). But it shows that there aren't "algorithms" invovled.
Message-ID: <4jrlnu$bfo@charm.magnus.acs.ohio-state.edu>
Message-ID: <4i3vpu$8lb@news.fsu.edu>
Message-ID: <4jallr$4dd@news.cic.net>
Message-ID: <1996Apr03.1146.scs.0004@eskimo.com>
Message-ID: <graphics/colorspace-faq_828120041@rtfm.mit.edu>
Message-ID: <chinese-text/faq_828569567@rtfm.mit.edu>
Message-ID: <4k2ug9$avo@redwood.northcoast.com>
Message-ID: <tv/the-prisoner/part1_828814601@rtfm.mit.edu>
Message-ID: <4julvm$n08@uunews.computize.com>
Message-ID: <games/corewar-faq_828569567@rtfm.mit.edu>
Message-ID: <folklore/afs-faq_828814601@rtfm.mit.edu>
Message-ID: <privacy/info-source_828718022@rtfm.mit.edu>
Message-ID: <4jt17f$bo1@mozz.unh.edu>
Message-ID: <4jrtpd$ngd@panix.com>
Message-ID: <1996Apr03.1143.scs.0001@eskimo.com>
Message-ID: <4jufij$d6o@mozo.cc.purdue.edu>
Message-ID: <4jrkup$o5o@cerberus.umd.umich.edu>
Message-ID: <4k4lni$8ik@apollo.isisnet.com>
Message-ID: <linear-programming-faq-1-828380411@cray.com>
Message-ID: <nonlinear-programming-faq-1-828380429@cray.com>
Message-ID: <usenet/software/nn/faq/part1_827545503@rtfm.mit.edu>
Message-ID: <4k175q$33p@newstand.syr.edu>
Message-ID: <4k3g1b$hns@tribune.usask.ca>
Message-ID: <usenet/software/nn/faq/contents_827545503@rtfm.mit.edu>
Message-ID: <rgra_faq_828647169@BRONZE.LCS.MIT.EDU>
Message-ID: <4jt146$bat@mozz.unh.edu>
Message-ID: <physics-faq/measurement-in-qm-1-828374162@mtnmath.com>
Message-ID: <4juf76$qvf@lace.colorado.edu>
Message-ID: <comp-graphics-gnuplot-faq_828584582@fg70.rz.uni-karlsruhe.de>
Message-ID: <4jufg7$cv6@mozo.cc.purdue.edu>
Message-ID: <nlp_828666593@cs.columbia.edu>
Message-ID: <digestfaq_828714006@ferret.ocunix.on.ca>
Message-ID: <star-trek/stardates_828569567@rtfm.mit.edu>
Message-ID: <tv/babylon-5/faq_828814601@rtfm.mit.edu>
Message-ID: <tv/babylon-5/setting/season2_828814601@rtfm.mit.edu>
Message-ID: <tv/babylon-5/setting/season3_828814601@rtfm.mit.edu>
Message-ID: <tv/babylon-5/news-schedule_828814601@rtfm.mit.edu>
Message-ID: <tv/babylon-5/setting/season1_828814601@rtfm.mit.edu>
PP> And why couldn't we use more than 32-bit of hex for a serial number?
We could, but that'd break all the stuff out there now.
PP> We're supposed to compare to whole thing as a string, there's not a
PP> big gain by storing the address and the serial number separately... To
PP> keep it the same length and augment the number of bit and move the
PP> limit upward a little, we could use base-36 numbers... That gives us
PP> about 650 times the number of serial number possible...
Just go with the Internet format <some_string@adress>. One frequently
used format for "some string" is like this:
yyyymmddhhmmss_TZstr_userid_task#_seq#
or
unix-timestamp_userid_task#_seq#
That pretty much covers it. You don't *need* to keep the sequence
counter, because it only matters for that particular task/window for
that particular user at that particular date and time. As soon as the
seconds change, you can start the counter over. No possibility of
duplication unless you completely screw up the system time, and in that
case you deserve what happens.
... A fool and his money soon become a sysop!
--- Blue Wave/DOS v2.30
* Origin: Shadowshack (1:105/51)