Great Circle Associates Majordomo-Workers
(December 1997)
 

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

Subject: Re: 1.94.4 resend
From: Jason L Tibbitts III <tibbs @ hpc . uh . edu>
Date: 06 Dec 1997 15:10:14 -0600
To: majordomo-workers @ greatcircle . com
In-reply-to: Walt Haas's message of Sat, 06 Dec 1997 10:10:38 -0700
References: <199712061710.KAA11446@ski.foo.bar>

>>>>> "WH" == Walt Haas <haas@xmission.com> writes:

[internal docs]

WH> In order for me to write something like that I would need to invest a
WH> lot of time in reading your existing documentation and the code.

Well, I need to do it anyway.  It's not really a stretch to fault me for my
development practices because I like to implement something a couple of
different ways to see how it works, then document it, then forget about it,
then read the documentation and see if the implementation seems dumb, then
start over.  So its often in the documentation process that I've stepped
back far enough to see where I need changes.  I suppose many people would
say that I need to document it first, then write it, but I think that's
like writing the dissertation before doing the research.

WH> That might actually be good, since it would give me a little different
WH> perspective than you now have - forest vs. trees type of distinction.

I'm sure that someone reading the stuff now will find lots of dumb things
and bad design decisions.  That's life.  It's also why things are
pre-alpha; I want to implement better ideas as they come about.

WH> At my site there are really three audiences for documentation, each
WH> with different needs.  So I try to write for each of them.

[...]

I think MLMs are part of a rare group of software this is true, in there
with operating systems.

I can't fault any of your categorization.

WH> As far as I can see, the most immediate need is to make Mj2 usable by a
WH> critical mass of people.

Well, for me the truly immediate need is to get things to the point where I
can run my lists, because running real lists is the only way to get real
testing.

WH> My heart wants to do the latter and my stomache is alarmed at the
WH> resulting low income :-)

That seriously differentiates us; I do this all in my free time and don't
expect to ever make any money off of it.  That doesn't mean I have a
different view of quality, just that there's nothing urging me to work on
the old code.  But honestly, today the smart move is to work on 1.9x and
keep the ideas around so that they will be in 2.0.  I just don't think that
it's going to be production quality in the near term.

WH> Presumably if we have a list member who needs help, all we know about
WH> them is that they can read basic email, so that's the preferred medium.

My grand design is to include in each interface an integrated help system.
For the email and shell interfaces this is just the help command, but for
the web interface this is a navigable document.  The trick is making it
come from only one place, so we don't have to maintain a pile of divergent
documentation.

WH> I've had mixed results from the FAQ files on the net.  For one thing,
WH> the FAQ for any real system tends to get huge, and less and less of it
WH> becomes applicable to any given problem.

That only means it needs to be broken up and/or refined, and perhaps that
the interface to the FAQ needs to be improved.

>> Heck, we could even have a FAQ search within the code itself.

WH> That could potentially be valuable for Audience 3, the Perl hacker.

I only meant that the user could submit a search string and the system
could respond with FAQ items that might apply.

WH> ... and I'm no longer the one to ask, either.  I'm really disappointed
WH> that I didn't get more feedback from the majordomo-users list.  By now,
WH> with the second alias file and mail spool, my installation is too
WH> different for the average site to get any benefit out of my document.

That's always going to be a problem.  2.0 generates the alias entries
itself [1] so I hope to avoid policy skew with that one, but the details of the
local MTA are just going to get in the way.  There isn't really any way
around it.  (Well, besides having Majordomo be the SMTP server...)

WH> I got thru make install but haven't set up a list yet.  Thought I had a
WH> bug for you but it was just Data::Dumper installed wrong.

Sigh.  Relying on external modules is clearly the way to go, but there's no
doubt it's going to cause problems in the long run.  syslog.ph is going to
be the biggest pain now that we've worked around the setuid problem.

WH> Wish we could get Mj2 over the hump ....!!

Wish there were 24 or so extra hours in each day _and_ that I could win the
lottery (without playing, no less).

 - J<

[1]
(/home/tibbs/mj/2.0-snap/bin/mj_shell -d morpheus.hpc.uh.edu -p gonzo createlist=nocreate GLOBAL)
Please add the following lines to your aliases file, if they are not
already present.  You may have to run the "newaliases" command afterwards
to enable these aliases.

# Aliases for Majordomo
majordomo:       "|/home/tibbs/mj/2.0-snap/bin/mj_email -d morpheus.hpc.uh.edu"
owner-majordomo: tibbs@morpheus.hpc.uh.edu,
majordomo-owner: tibbs@morpheus.hpc.uh.edu,
# End aliases for Majordomo


References:
Indexed By Date Previous: Re: 1.94.4 resend
From: Walt Haas <haas@xmission.com>
Next: restrict_post and Reply-To addresses
From: Nick Perry <nick.perry@amulation.co.uk>
Indexed By Thread Previous: Re: 1.94.4 resend
From: Walt Haas <haas@xmission.com>
Next: approve to override valid_addr?
From: Bill Houle <Bill.Houle@SanDiegoCA.NCR.COM>

Google
 
Search Internet Search www.greatcircle.com