This kind of idea has gone around before, and there are lots of ideas on
the best way to do the nagging.  My taste would be not to have a
notifier pop up, mainly because this is a little too strong of a nag. 
Also, it bothers me to stop as soon as every class has one comment, but
we also need to worry about getting old comments updated.

My favorite "nagging" approach would be to include the comment in the
class definition window.  That way people see it all the time.  It would
be just like with methods: whenever you browse to a method/class, the
second thing you see is the method's/class's comment.  I ran my image
for a while in this way and I found it very smooth.  For once I actually
knew a little about the class comments in my image.  Somehow I never got
in the habit of manually clicking "?".

There's a close tie to the design of hypertext and hypertext systems,
e.g. WWW sites.  If you require people to explicitly click a link to
check out what something is, then they need a firm reason to do the
click.  It is better to give a little bit of a hint of what is behind
the link.  For a picture, give a thumbnail.  For an article, give the
title and the word count.  Here in Squeak we have an issue that the link
to the comment gives no summary of what is behind the click.

Anyway, I posted some code to simply dump the comment into the class
definition window.  The list's reaction was lukewarm.  It seems that
everyone has their own favorite way to nag.  :)


-Lex