Pat, I've been catching up with about ten days of TD (out of town and
then a cold... :-) ), and was wondering what was up, and didn't see
you discuss it (I could have missed it ...).
I got about 6-10 issues with no 'number' information in the subject,
and each was exactly 30K long. These were about numbers 184-190
(from the inside info), I think.
Then/and, other issues before that had the sender in various formats.
More issues from being spam-overloaded or something else?
Thanks for your illuminating info, as always!
Lee Sweet
Datatel, Inc.
Manager of Telephony Services
and Information Security
How higher education does business.
Voice: 703-968-4661
Cell: 703-932-9425
Fax: 703-968-4625
lee@datatel.com
www.datatel.com
[TELECOM Digest Editor's Note: Ah, so you noticed ... well, I at first
thought that maybe John Levine, who administers the mailing list
through majordomo had fallen asleep at the switch(board) ... although
I did think it a bit unusual that my volume of hate mail had fallen
off a little (although the amount of spam and viri had stayed
constant or maybe increased a little.) Then I discovered that about 6
or 7 issues (184 through 190 I think) had not made it though the
sendmail here at massis. I do (at least) two things here: a script
of mine, from years back when I still had all my marbles intact,
called 'posterd&' (the ampersand means to run in the background) which
in turn calls 'nntpxmit' (a process which posts comp.dcom.telecom on
several sites at the same time, message by message) and 'hypermail'
(a script which posts the individual messages in each issue 'usenet
style' message by message on http://telecom-digest.org/TELECOM_Digest_Online
_AND_ I also run 'send-telecom' (which itself calls up a background
job called 'send.now', the purpose of which is to (1) make up the
latest-issue.html file on the fly and (2) preserve a copy of the
Digest in our archives in numerical order and last but not least,
(3) invoke [or would you say PROvoke] an instantiation of sendmail to
get this out to the mailing list readers, with a few clever tricks
out of sendmail which I am allowed to use as a priviledged user
at MIT [such as the -f flag allowing sendmail to refer to me as
'editor@telecom-digest.org' rather than just as 'any old sap user@
lcs.mit.edu'. [Quite aside, but now and then some other site somewhere
on a spam-fighting tangent discovers that our sendmail here allows me
to make 'those adjustments' and it kicks my stuff back or shreds it
undelivered, whatever].
In the process of doing these two jobs, I sometimes grow impatient,
since hypermail -xp takes fu---- forever it seems, now that I have
those Google ads in there, and the larger my stash of messages gets,
the more ads I get people to look at, etc. So my diseased brain, in
its wisdom thought of an 'improved' way of handling things: I would
invoke posterd& and after seeing it get happily on its way doing its
job I would then invoke send-telecom, letting them both run at the same
time, while I went to have a cigarette and feed the cats. Ah, but this
new scheme of mine, this shortcut led straight to hell, an express
lane so to speak; it failed, miserably, as diseased brains are want to
do; the sendmail (portion) of the send-telecom job got broken, because
sendmail could not find a couple files it insisted on having, since
the earlier script (posterd&) had not yet had a chance to create them
for sendmail to find. _Not all_ scripts running in the background
which invoke sendmail bother to tell the invoker when something goes
wrong, so I sat here blissfully assuming that things were going along
fine.
I still have not figured it all out yet, although transparently 'it'
works the way it should from the reader's perspective. You cannot write
a Unix script without lavish commenting to remind yourself what you
did [and no _real man_ ever bothers to document his scripts] then go
back twenty years, two heart attacks and one brain aneurysm later and
understand what you wrote or how to fix it if it gets broken. By purely
hook and crook, I found out one thing (after I had manually forced
those six or eight issues you got all at once through the sendmail
slot at the postoffice): the regularly scheduled issue got through
okay! But I had done it _exactly, precisely_ as I had done it in the
past when posterd& and one of its components hypermail -xp only took
a few minutes to run instead of the twenty minutes it takes now with
all those God Damned Google ads that have to be on each page it
creates. So ... if I do it in _this_ order (posterd& _then_ send-telecom)
by George, John Levine gets it, the readers get it, etc. So, there
has to be one or more files _in common_ to both scripts which send-
telecom needs (but posterd& has not yet created or has created but
is working on them when send-telecom comes looking for them or maybe
posterd& had the audacity to rm them while send-telecom was using
them or looking for them, etc), I dunno. I have the scripts back to
where they mostly work, but I still have to do some manual 'tidy up'
afterward on them. Ah, if we could have only known twenty (even ten!)
years ago what we know now about mailing lists, interaction with the
web and stuff like that. Maybe someday my hero will come along and
work all the inconsistencies and bugs out of my scripts so I can do
this in a modern way. I know I sure cannot do all that any longer.
Enjoy it and use me while you can. Someday I won't be around to get
kicked any longer. PAT]
Lee Sweet (lee@datatel.com)