Re: Hyphenation (was Re: A suggested tag)

F. E. Potts (fepotts@fepco.com)
Sat, 19 Apr 1997 18:02:31 -0600


Date: Sat, 19 Apr 1997 18:02:31 -0600
From: fepotts@fepco.com (F. E. Potts)
Message-Id: <97Apr19.172416mdt.18434@gw2.fepco.com>
To: papresco@calum.csclub.uwaterloo.ca
Subject: Re: Hyphenation (was Re: A suggested tag)
Cc: www-html@w3.org

On Sat, 19 Apr 1997 14:48:04 -0600, Paul Prescod wrote:
> As the anal person who responded to Carl, I should point out that I
> did not state that HTML should never have any presentational markup
> in it.  Actually I think that the occasional word break marker is a
> lot easier then trying to tag the semantic meaning of words with
> different hyphenation points in different contexts. But recognizing
> that the line must occasionally be crossed is different from arguing
> that there is no line: even a fuzzy one, as Carl wants to do. Yes the
> line is occasionally fuzzy. Yes the line is worth crossing for
> convenience sometimes. But yes, there is a line. The distinction
> between structure and presentation is not something invented on this
> list to complicate discussions and confuse users. I don't think it
> was your intention to support that view.

My comment was not directed at you, but I think you understand that.
:-)  And yes, the line does exist: we all have to deal with it now and
again, trying to keep things as clear as possible while at the same
time not backing ourselves into a corner based on theory rather than
real-world needs.

It was Carl's seeming position that there was no line that bothered
me.  Our entire field is built on this separation, and how to deal with
it gracefully.  It is, in fact, where our power lies.

> As far as your example, it doesn't seem particularly convenient to me
> to change your document by adding in these "..." elements when you
> could do it in one place in your style sheet. If your style sheet
> language is weak then this is the best choice, but with DSSSL it
> would be trivial.
>
> (if (not (string=? (first-child-gi) "TITLE")) (make rule
> orientation:  'horizontal ) ... )

A FOSI is never convenient; I must have done something bad in a
previous life to have been--no matter how temporarily--condemned to
using them.  None of my print-outs (except for snail letters) are used
for anything but work documents, since final book production is turned
over to specialists, so I use what I have available.  Essentially, the
FOSI I use for editorial manuscript production is a sample that came
with my current SGML tools that I modified for my use--I have neither
the time nor the inclination to become a FOSI expert and build a
high-quality application that can be modified easily: especially since
the FOSI is (IMO) a dying tool.  Hopefully, in the near future a good
DSSSL engine will become available that will work with my Unix tools,
and at that time I will happily make the switch (that is, if it
provides output as beautiful as my FOSI/TeX engine does).  But my main
concern is, as always, with my content and the quality of my SGML; the
various forms of presentation (paper, web, etc.) I deal with after the
fact.  Which, of course, is the way it should be.

> (I'm always hesitant to post DSSSL code, because the fact that it can
> be used to do things that take a couple of minutes thinking serve as
> "evidence" that it is hard to use. On the other hand, simple examples
> may serve to convince people that even the advanced features of DSSSL
> are not that hard.)

Quite the contrary.  I have no hands-on experience with DSSSL, and the
more examples I see, the happier I am.  A few minutes spent thinking is
always better than a few hours spent in "grunt" work at the keyboard.

> The nice thing about the purist approach is that it usually saves
> work in the end. With DSSSL, it is two minutes of typing rather than
> 20 of hunting around for instances of sections with no titles. I'd
> guess you weren't using DSSSL though.

Go ahead, rub it in, Paul. <g>

> I'm also not sure why you would manually type &bull; &bull; &bull;
> rather than generating the bullets. What if you want an ordinary
> horizontal rule tomorrow? Is &bull; a statement of opinion about your
> editor or the weak stylesheet language that you were stuck with? =)

Don't tempt me. <g>

As to the character-entites, using them allows me to use any dingbat I
wish, even my edi-- oops, there I go. :-)  Actually, because of the
design of the book FOSI I have, it turned out to be simpler to just
drop <separator> and its attributes into the DTD and make a minor FOSI
modification.  My first choice, as I mentioned, was generated text; but
it turned out that that would take far too much time to implement.

> To summarize, I have no problem with the occasional presentation
> markup thrown in where the generated version would be a real pain:
> for instance those rare ambiguous hyphenations, but the distinction
> between structure and presentation exists and doing things the
> structural way will usually save time and money. The other benefits
> are often gravy.

Very true: I couldn't agree more.

-fep

--
fepotts@fepco.com
http://www.fepco.com