Great Circle Associates Majordomo-Workers
(March 1997)
 

Indexed By Date: [Previous] [Next] Indexed By Thread: [Previous] [Next]

Subject: Re: [Fwd: Re: Syntax for a 'set' command?]
From: Jason L Tibbitts III <tibbs @ hpc . uh . edu>
Date: 21 Mar 1997 15:37:55 -0600
To: Chael Hall <nowhere @ chaos . taylored . com>
Cc: majordomo-workers @ greatcircle . com
In-reply-to: Chael Hall's message of Fri, 21 Mar 1997 16:16:40 -0500 (EST)
References: <Pine.BSF.3.95.970321160723.906K-100000@chaos.taylored.com>

>>>>> "CH" == Chael Hall <nowhere@chaos.taylored.com> writes:

CH> What kind of low-traffic list is this?

Huh?  I have no idea what your talking about.  If you take exception to the
numbers, then you probably didn't notice that they were just examples.

The point is that there are several subscription classes and we need a way
to set them.

CH> My X-Files list rarely gets under 5 digests per DAY.  I'd like to offer
CH> bigger digests, but the AOL subscribers couldn't receive over 22k
CH> digests without "downloading" them separately.  They complain, I
CH> modify...  As long as the feature is configurable per list, I think
CH> it's a great idea.

That's a lot of traffic.  I originally had the idea for personalized
digests where the user could choose the limits, but I think that's going
overboard.  I figured two or maybe three different digest classes was good
enough.

CH> You need the size limit in case there's a whole bunch of mail in a
CH> given day.  Without it, the poor list members could suddenly receive a
CH> 300k digest which will undoubtedly bounce from many recipients.

I really don't know how the digest triggering mechanism works.  I figured
that in the most luxurious implementation there would be four different
triggers (size, messages, time, oldest message) and one limit (minimum
message count) that would combine in some way.  I haven't really thought
about it.  Perhaps minimum message count would override time, so that you
don't send out a digest with only one message in it (perhaps because just
before midnight you kicked off one because of the size limit).  If someone
wants to write some kind of spec for this I'd be glad to follow it.

The point is that digest can say "deliver this message to this class of
subscribers, except for these users" and the delivery engine takes care of
it.  This works already.  Building up the digests is something that needs
to be done.

And then there's the issue of MIME vs. traditional format digests.  That
should arguably be a per-user thing, too.  Ugh.

CH> One thing that would be good is to send partial digests when the
CH> subscriber changes from digest to instant.  ListProcessor used to do
CH> that...  That was one of two features that I liked about it that Mj
CH> doesn't do.

That should be pretty easy to figure out, once the digest gets written.
What was the other feature?  (Well, keep it in mind; I'm not really on a
feature hunt right now as my TODO list is huge already.)

CH> Which reminds me...  Do you support something like this:

CH>      approve password command1 command2 ... commandn end

CH> The approve applies to all commands after it (until perhaps another
CH> approve that overrides it?)  I hate typing passwords on every line.

default password blah
default list hurl
command1
command2
...

The defaults stick around.  (Default list works just like -l does now; in
fact calling mj_email with -l just does an implicit "default list
$opts{l}".)  You can unset them with

default password

to send unapproved commands.

CH> The syntax is good.

OK.  That's one.  But what about the flag settings and naming of the
digests?  Should we try to copy the flags (ack/noack) when they coincide in
purpose?  Some of the listserv flags seem to have pretty unintuitive names,
though.  (Sorry, no example at this moment.)

I'm starting to lean towards

set list digest (defaults to something the list owner sets)
set list digest-large (-medium, -small)

instead of "set list digest large" because it makes the parsing a bit
easier.

 - J<




Follow-Ups:
References:
Indexed By Date Previous: Re: [Fwd: Re: Syntax for a 'set' command?]
From: Chael Hall <nowhere@chaos.taylored.com>
Next: Re: A few patches against mj 1.94.1
From: Vince Skahan <vds7789@hps12.iasl.ca.boeing.com>
Indexed By Thread Previous: Re: [Fwd: Re: Syntax for a 'set' command?]
From: Chael Hall <nowhere@chaos.taylored.com>
Next: Re: [Fwd: Re: Syntax for a 'set' command?]
From: Chael Hall <nowhere@chaos.taylored.com>

Google
 
Search Internet Search www.greatcircle.com