Subj : Something new...
To : Richard Webb
From : mark lewis
Date : Thu Oct 15 2009 06:48 pm
RW> <snip again>
RW> AS for routing, squish or his mail processor would handle
RW> that as Bink is a static mailer. SAme inbound/outbound dirs though
RW> pointed to by all versions.
ml> hummm... therein lays the problem... can tossers handle multiple
ml> routings for static mail bundles?? FD can because it is dynamic in
ml> building its outbound stuffs but if a BSO style tosser cannot build
ml> multiple ?lo files indicating different "status'" for the same
ml> static outbound bundles for separate nodes, this may be a problem...
RW> not an option. Instead, what you do is process the
RW> nodelist. EXample, you've got a node you regularly connect
RW> with and crash packets to. This node has both pots and your other
RW> common scheme. YOu don't want to dial pots to send
RW> those crash bundles. so, you replace that node's pots phone no
RW> with unpublished using your favorite nodelist compiler.
RW> Need to change your mind and dial him pots? Grab the right
RW> phone number from the raw nodelist as distributed, recompile
RW> nodelist, you're happening. Other scheme comes back live,
RW> switcharoo your nodelist compiler control file back,
RW> recompile.
ugh... i really do like FD's nodelist compiling stuff... all FD does is
generate b-tree indexes on the nodelist and then if there are overrides for any
nodes, those go into the b-tree index so they are used instead of the default
nodelist data... with FD being able to use environment variables for nodelist
entries, one simply replaces the phone number with the environment variable and
then the mailer (FD) looks to the environment variable for the number to use...
since each node maintains its own environment settings, one node can have
set foo=1-800-555-0100
and another node can have
set foo=some.domain.whatever
where "foo" is the EV (environment variable) assigned for the override for a
node's phone number...
if i have stuff for a node that has both POTS and FTN over telnet, and i want
to "crash" (i use immediate instead of crash) what i have waiting for them via
POTS, i simply go to one of the POTS nodes, look at that mailer's outboud queue
and change the flavor of that destination system's bundle(s) from hold to
"normal", "crash" or "immediate"... all the other nodes still retain their
settings for that destination... the one mailer node then does the connection
and signals the others when it is successful so they will rescan and rebuild
their outbound control files... i don't have to shut down all the mailers, take
the few minutes to recompile the nodelist and restart all the mailers...
RW> Most nodelist compilers that create a nodelist binkley can
RW> use will let you do the above quite handily iirc.
i'm sure ;)
ml> i guess FD's dynamic stuff is similar to having it build its own ?lo
ml> files but those ?lo files are also specific to the node as well as
ml> to the destination system... FD uses the file extension to denote
ml> the node number the file is for and the content of the file denotes
ml> the addresses and "flavor"...
RW> YEp, similar. YOur tosser and other utilities build the
RW> *.*lo files. There will be a file in my outbound when I
RW> write this message created/updated by squish with a file
RW> name of 0e32000c.clo detailing paths and filenames to be
RW> sent to you.
right... and there's the main difference... there's only one ?lo file for all
the mailer nodes to see and use... it cannot be a hlo, flo and clo all at the
same time so that individual mailer nodes can act on it the way they need to...
RW> so, if you and I had multiple connection schemes between us
RW> such as telnet and pots and I were running multinode I'd
RW> possibly want to prefer the telnet connection. But, in my
RW> case maybe I"d prefer the pots, only use the telnet if there were
RW> trouble on lines between us. IF I want to prefer the
RW> telnet I change your pots entry to unpublished so that bt
RW> pots doesn't try to dial you.
right... that is one method... it can also be done like this with FD but it is
much easier and simpler to simply set that mailer node's routing table to hold
that mail during that scheduled event... so these mailer nodes see that mail as
"hold" and those see that exact same stuff as "normal" (or "immediate" or
"crash" or whatever) and each can act on it separately...
RW> But, if I want to prefer the pots as primary I"ll want to
RW> change your nodelist flags entry, again doable with my
RW> nodelist compiler to not show the telnet capability. Or, I
RW> can leave the nodelist alone, and the mode that gets there
RW> first transacts a session with yours. Meanwhile, if
RW> assuming I"m running other processes since I"m running
RW> multinode those other processes are aware of which one's
RW> talking with you and won't touch mail bundles to/from your
RW> system.
yup... i've had times where i've had more than one POTS node pick up the line
and dial to the remote while at the same time, more than one of the internet
nodes also went to connect to the same destination... in that case, it is the
luck of the draw as to which one gets the first connection and makes the
transfer ;)
RW> <snip again>
RW> Bob Juge would have been your guy for that as he was doing
RW> it. Also the other MR. LEwis might be of some assistance to SEan
RW> in this regard. THink he's running bt.
ml> yup and yup on both accounts...
RW> When I helped a friend get it all going I referred to things I"d
RW> seen in this echo quite a bit, but that was way back in
RW> the day <g>.
as was with many echos in fidonet, there was a huge amount of information on
almost any subject that traveled between all of our systems... it really is sad
that folk have folked to the internet for eye-candy without the real meat of
the meal...
