Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!bbn!gateway!RELAY.NSWC.NAVY.MIL!dsill
From: dsill@RELAY.NSWC.NAVY.MIL
Newsgroups: comp.emacs
Subject: webster and functions on words
Message-ID: <37520@bbn.COM>
Date: 21 Mar 89 16:37:34 GMT
Sender: news@bbn.COM
Organization: BBN news/mail gateway
Lines: 51

Yesterday I posted a version of webster that provided the current word
as the default word to be spelled, defined, or ended.  Also yesterday,
Dirk Grunwald posted a version of webster that provided webster-word
and webster-define to spell or define, respectively, the word to the
left of point.

There is an inconsistency in the way Emacs functions on words work.
Some, like ispell-word or webster-word, operate on a word in the
current buffer.  Others, like apropos or webster, read their argument
from the minibuffer, after prompting.  Some of the latter, like
unix-apropos and webster as I modified it, provide a default response
of the current word.

Personally, I think it's better to have one function that can operate
on either the current word or one read from the minibuffer than to
have two separate functions.  The former has the advantage of showing
the operator exactly what word will be operated on if the default is
taken.  The only thing that could be considered a drawback of this
approach is that it requires the operator to confirm his choice by
pressing return.

What's the consensus: two separate functions, one operating on the
current word and one reading from the minibuffer; or one function
reading from the minibuffer providing the current word as a default? 

A related problem is that these functions determine the current word
differently.  Most, like ispell-word, use the regular expression \\w
to find the complete word to the left of point.  Others, like
webster-word, use the partial word from point to the beginning of the
word to the left of point.

Perhaps the function ``interactive'' could be modified to include a
code, say "w", meaning "the complete word to the left of point".  But
that would only solve half the problem.  There would still need to be
some way to handle defaults.  How about a general default mechanism
added to ``interactive''?  Something like:
    (interactive "wISpell word (%s): ")
    (interactive "aDescribe function (default %s): ")
which would result in:
    ISpell word (prefrobnicate):
    Describe function (default interactive):

Comments?  Flames?  Implementations?

================
Above opinions are mine.

"The ultimate metric that I would like to propose for user
friendliness is quite simple: if this system were a person, how long
would it take before you punched it in the nose?"
					-- Tom Carey