You probably remember the program I had mentioned earlier which I had said I
would use to determine the nearest peer to the invoker of a SUBscribe command,
wherever he may be on the net. For those of you who remember it, this is the
*very* program that Mike did not want me to waste time on :-) Well, I had said
that it would have to be able to generate the list of all nodes and their dis-
tances to any given source node in less then 0.30 secs TCPU on my 4341-2, or it
would not be used at all. Help yourself :-)
*-----------------------------------------------------------------------------*
kl frecp11 bearn blekul11 tamvm1 uiucvmd clvm uga
BEARN 3
BLEKUL11 4
TAMVM1 13
UIUCVMD 9
CLVM 9
UGA 11
R; T=0.23/0.25 22:33:04
kl bitn*c frecp11 tamvm1
FINDNODE error
R(00101); T=0.03/0.04 22:33:47
kl bitnic frecp11 tamvm1
FRECP11 5
TAMVM1 9
R; T=0.23/0.25 22:33:51
kl tamvm1 bitnic
BITNIC 9
R; T=0.23/0.24 22:34:00
*-----------------------------------------------------------------------------*
Close call, but I did win :-) The same module in REXX takes 2 minutes to produ-
ce the list of 1 to 8 links distant nodes, and then you really can't help doing
HX....
(this is a temp version -- the syntax is "KL sourcenode list-of-nodes-you-want-
stacked) The whole list is built in storage whatever the arguments, so it takes
very little additional time to generate the whole list (kl frecp11 *). I'm glad
to announce that this program will now be used to generate the NODES FILE from
LSVBITGN, and LSVGNODE will become obsolete. It particular you will not have to
stop the server to run GENROUTS. There will be a new command, "NODEGEN" (or
something similar) to perform all the BITEARN NODES based generation process
automatically from within the LISTSERV machine -- that will save you a lot of
time and will save ME a lot of mail :-) A nasty side-effect: LSVBITGN now takes
200 secs TCPU on my machine instead of 160, but then you run it only once a
month. You need an additional 15k of datafile, but I think this is no big deal.
I'll test that darned little module until it breaks or 1 hour has elapsed, and
then I'll modify SUBSCRIBE to look at the "Peers=" keyword in the list header
and forward the command if needed. I also plan to implement a MOVE command
soon, we really need it: MOVE listname u@n <TO> newnode. This would forward
the complete set of distribution options, not just the name and [log in to unmask]
I also had to fix a problem with CARD which does not accept a filemode of "*"
(*groan*) thus causing an internal error.... So I guess I'll call this 1.4b.
*-----------------------------------------------------------------------------*
Here are the long awaited (??) results of the "dear sir" vote... :-)
1 (1) 'Dear sir'
1 (2) 'Dear sir or dear madam'
(2bis) 'Dear Sir Or Dear Madam'
1 (3) 'Dear User'
+2 (3bis) 'Dear LISTSERV user'
2 (4) 'Dear Networker'
(4bis) 'Dear Network fan'
(5) 'Dear person'
(6) 'Hello'
4 (7) 'IAT8187 JOB 4531,NEWUSER PR=15 CL=S AWAITING GREETING(P)'
1 (8) 'Dear new list member' (ADD)
'Dear old list member' (DROP)
'' (Notes sent to list owners)
1 (9) 'Dear interested user'
2 (10) ''
1 (11) 'Dear subscriber'
'Dear list owner'
1 (12) 'Greetings friend and neighbour'
1 (13) 'To whom it may concern'
1 (14) If LSVBITNIC_staff(userid) Then greeting = 'FYC'
Else greeting = 'Dear sir or dear madam'
(This does not include my own vote)
The winner is therefore item number 7, IAT8187 JOB 4531,NEWUSER PR=15..... I
therefore decided that my vote would count as 25 points ( ;-) ), and I voted
for item (15), which is what you see in the V1.4 $DEFAULT MAILFORM (although
items 14 and 4 were tempting). Morals: never put something on a vote, people
will always react like the Schtroumpfs in that cartoon where they all said:
"Moi, je vote pour moi!!!" (translation: "*I* vote for *myself*!!!") :-)
Eric
|