From ...
From: Erik Naggum <erik@naggum.no>
Subject: Re: LOOP and patterns (Re: Hash to list?)
Date: 1999/12/13
Message-ID: <3154094846488075@naggum.no>#1/1
X-Deja-AN: 560150729
References: <gmln74hzgz.fsf@ola.hex.fi> <87k8mo9is1.fsf@senstation.vvf.fi> <gmd7sghxbv.fsf@ola.hex.fi> <2hbt80kpij.fsf@dslab7.cs.uit.no> <3153828708572466_-_@naggum.no> <2hd7scj56p.fsf@dslab7.cs.uit.no>
mail-copies-to: never
Content-Type: text/plain; charset=us-ascii
X-Complaints-To: newsmaster@eunet.no
X-Trace: oslo-nntp.eunet.no 945106046 7155 195.0.192.66 (13 Dec 1999 17:27:26 GMT)
Organization: Naggum Software; +47 8800 8879 or +1 510 435 8604; fax: +47 2210 9077; http://www.naggum.no
User-Agent: Gnus/5.070099 (Pterodactyl Gnus v0.99) Emacs/20.5
Mime-Version: 1.0
NNTP-Posting-Date: 13 Dec 1999 17:27:26 GMT
Newsgroups: comp.lang.lisp

* Frode Vatvedt Fjeld <frodef@acm.org>
| Well, I disagree with you here.  I think "(hash-table-list x)" is a lot
| more readable (and not the least bit weird) than the loop expression,
| which I find about equally readable to the "plain" lisp form.  And, more
| importantly, the time-to-convince-me-it's-correct factor is also much
| better.

  well, I would _not_ understand what a function called HASH-TABLE-LIST was
  doing without reading its documentation or source code, and that I would
  have to repeat every time I found it in a new package or project.  the
  reason is quite simple: there are so many lists that could be generated
  from hash tables that none of them merit being the only one thus named.

| I'm a bit confused: My impression after reading OnLisp and other books is
| that I'm encouraged to write and use precisely this kind of "utilities".
| Do you disagree with this?

  actually, yes.  naming utility functions is a very hard problem, and they
  should therefore not be multiplied lightly.  when there is clearly a
  diversity of needs, it is better to be a little more verbose than to
  pollute the name space with thousands of functions.  the C++ way to do
  this is to talk about patterns to discourage people from creating tons of
  one-call functions.  the Lisp way is to create sub-languages and macros
  that cover _multiple_ needs well.  I actually think LOOP succeeds in
  this, but it suffers in my view from not creating an environment with
  local macros instead of the positional "lexical keyword" stuff that I
  have come to dislike in other languages.

#:Erik