[plt-scheme] Re: to define, or to let (last try)

From: Bill Richter (richter at math.northwestern.edu)
Date: Sat May 1 23:31:17 EDT 2004

Thanks, Shriram.  I think what's needed is 1-way pronouncements from
the PLT team: you, MF, Matthew, Robbie...  I think the problem is that
unmoderated discussions can really chew up the time & emotional energy
of the PLT maintainers.  So I asked you & MF for final pronouncements,
and I was satisfied, but one person I talked to disagreed with me
about what MF meant!  So can you comment on my pronouncement:

 PLT Scheme is very interested in computer error messages.  So we
 didn't like the R5RS paradigm, where the coder was responsible for
 checking that the code was independent of order of evaluation.  The
 interpreter can't check eval order dependence statically, and it
 would be hideously expensive, if not impossible, to check it at
 run-time.  So we moved to strict left->right eval order.  It's a new
 paradigm, and we understand some folks don't like it.  Some folks
 think that humans were doing a good job checking eval order problems,
 and that the old paradigm encouraged functional programming.  But
 we're going this way.  If you want to argue, take it to c.l.s.

Back to you, Shriram:

   I don't dispute that some very smart people seem to be getting
   something out of this discussion.

I don't think I'm smart, and I'm certainly not much of programmer, but
I got a lot out of the discussion.  Anton even taught me some LC_v.

   Indeed, as far as I can tell, every single one of the people
   participating on the mailing list also posts on comp.lang.scheme.

Not me.  Northwestern doesn't have a news server anymore, so I haven't
read c.l.s. for 6 months.  But I'd dropped out a year ago anyway.

   If it's really an issue, I'd be happy to create a separate mailing
   list -- say plt-users or plt-community, or even plt-flames <-;.

Some reorganization is needed.  Most of the traffic on plt-scheme
looks like bug reports (which I've always submitted through Help
Desk).  Most of the rest of the traffic is about HtDP, but I didn't
know it was supposed to go here.  I've got an entry myself on the your
web site <http://www.htdp.org/2002-09-22/typos.html>, and it ends with
a mail engine called "The authors".  Does that send you to plt-scheme?

I posted that folks should write articles about how to use mutation in
the new l->r Mzscheme paradigm.  Where do want it?  If such articles
are posted on c.l.s., I won't see it, and they'll get followupped by
folks saying the new l->r Mzscheme paradigm is bad.

How about just making pronouncements on the plt web site?  Such as

"Mzscheme is no longer R5RS compliant, because our letrec = letrec*,
i.e. analogous let* instead of let."

And maybe say,

"if you don't understand our pronouncements, we'll tolerate some
questions on plt-scheme.  But if you want to argue, take it to c.l.s."

   You guys can yak away all you want on it.  I would even read it,
   probably in weekly digest mode.  If there's non-trivial interest in
   that, I can create it.  But please let's try to keep *this* list on
   topic.  Crying intolerance is, I believe, in this instance both
   unfair and misguided.

I didn't cry intolerance.  I thought you & MF put a stop to an
addictive discussion: I made a "final" post last Sunday, trying to get
out, and afterward, Paul Shlie wrote me a very nice letter, with the
Subject: please consider ending "Re: to define, or to let"
And I said, sure, let's quit, but Paul couldn't take his own advice :D
So we needed help kicking the habit, so thanks again.

   One of the things that seems to distinguish PLT Scheme is that the
   entire development team closely monitors this list and users get
   prompt responses to PLT questions.  It's important to us to
   preserve this invariant.  Wading through lots of non-PLT mail on
   plt-scheme makes this really hard.

Aren't you using procmail to run spamassassin?  You just had to stick
this in your .procmailrc:

* ^Subject:[ ].*to define, or to let

Maybe the issue is cluttering the list for the audience, not the

   The alternative, from a time-management perspective, is for the PLT
   members to read the list in digest mode.  That would greatly reduce
   response time, without any appreciable value to the userbase.

And let me apologize again, because I was doing digest, so I didn't
worry about being "indigestible".

Posted on the users mailing list.