Re: AWEB; Enhanced Document Encoding

[Sven Utcke <utcke@tu-harburg.de>]

"Nick Roberts" <Nick.Roberts@dial.pipex.com> writes:

> Sven Utcke wrote in message <36f04969.0@pfaff.ethz.ch>...
> |"Nick Roberts" <Nick.Roberts@dial.pipex.com> writes:
> |
> |> I have to agree that TeX -- to be completely blunt about it -- really creaks
> |> these days.
> |
> |Strange.  I usually refer to it as "TeX flys", especially when
> |comparing it to the 8086 4.77MHz I used to use it on.
> 
> Oh no!  I _didn't_ mean to start a flame war!  Help!  I feel compelled
> (maybe I shouldn't :-) to answer Sven's post, but from an objective point of
> view I think, not a hostile one.

Nah, you cna't objectively discuss TeX (or Emacs, for that matter)
without to acknowledge that they are simply the best...

> |I have never yet lost one of these two
> |bets:
> |a) given any moderately complicated formula, I can create a printout
> |   of it (using LaTeX) faster than the other guy can using WinWord
> |   (even though I'm batch-processing, while he's WYSIAYG).
> 
> Which (presumably) proves TeX is better than Word at formulae.  Fair enough,
> but Word, I suspect, is better than TeX at a great many things!

I suppose so.  Hmm.  I mean, it must be, after all, it's the
market-leader.  It's just I can't think of anything.  Do you think it
would qualify that WinWord can change the layout of your document
because you changed your printers resolution from 1200DPI to 600DPI
(and usually will)?

> Experiment (hypothetical).  Seat a young secretary in front of a computer
> with: (A) LaTeX and (say) noweb; (B) Word 7.0.  

They actually performed that experiment back in, uh, beginning of the
nineties, using Word 5.0 (I think) and LaTeX on secretaries who had
never before worked on a computer.  It turned out that the ones using
LaTeX were actually learning faster(*).  Our own secretary (well, from
about 3 years ago) always used LaTeX and could never get to grips with
Winword. 

> Ask the secretary to
> copy-type: (1) a letter to a disgruntled client; (2) some documentary notes
> on a program I am writing, including copious mathematical formulae; (3) a
> small literate program I have drafted.  Now consider the following
> scenarios: (A1) the secretary resigns in tears; (B1) all done in half an
> hour; 

I'm writing all my letters in LaTeX, and it rarely takes 30minutes.  I
think this comparison is only fair if you assume that the secretary
does know as much about LaTeX as she does about WinWord.  In WinWord
it would boil down to:
* This is how you load a document-template (or whatever they called in
  WinWord) and this is how you fill it in.
And in LaTeX
* This is how you load a document-template, and this is how you fill
  it in.
Not much of a difference, as far as I can tell...

> (A2) after offering a raise, it still takes two half-day computer
> courses and a week to complete; (B2) all done in five hours, with a certain
> amount of swearing, and a (cosmetically) nasty result; 

No, it won't work that way.  Chances are your secretary doesn't know
an integral from a sum, so in order to have her typeset any formula in
any software you'll have to spend a considerable amount of time
explaining what is what, and what it should look like.  Otherwise the
results will not be cosmetically nasty, but simply plain unuseable.
To many things in mathematics depend on the _exact_ look.  

What is more, once you have explainde the difference between a sum and
an integral sign, there is very little to stop your secretary from
typing \sum.  And since LaTeX knows a lot about what mathematics
should look like, she is actually quite likely to get good results.
Which is completely unlikely with WinWord, since WinWord requires you
to know how it should look like in order to make it look that way.

> (A3) after offering
> another raise, it takes a further two half-day computer courses and another
> week, but the result is fine;  (B3) no chance.  

I don't think your secretary will ever make a good job of a literate
program, no matter what tool she is using.

> In other words, horses for courses.

I agree that there is no reason why one shouldn't use WinWord for
letters.  As a matter of fact, it's perfectly fine for letters.
Nothing I couldn't do with LaTeX too, and probably faster (with a
little Help from Emacs), but it's absolutely adequate.  If it comes to
series letters (or whatever they are called.  Sending the same letter
to 200 people), WinWord might even be a tat easier.

The point is, while both WinWord and LaTeX are perfectly adequate for
writing letters, LaTeX will also produce beautiful books with lovely
maths.  WinWord will crash.  Or wont print.  Or will reformate your
entire 200-page book (this is of course hypothetical.  You can not
write 200-page books in WinWord if you need only even a moderate
amount of crossreferencing, pictures and formulas) if you changed your
printers resolution --- never mind changing the printer (say for 4
colour pages).  I've seen students do this on their first thesis-like
asignment.  Trust me:  I know of no student who has used WinWord for
his first assignment who has also used it for his second.  The
experience is just to horrible.

Not that I remeber why I felled compelled to make that point.
Probably something to do with restarting Windows all morning (trying
to get the very same network-card to work which worked flawlessly one
month ago.  All Windows ever does unfortunately is to tell my:  It's
plug and play, don't worry, all is fine.  Well, it's not...).

> Oh, and the secretary likes the paperclip best (apparently).

So do I.  Everything else about WinWord is a pile of ...., but the
paperclip is undeniably cute.

> |b) The typeset result produced by (La)TeX will look better (as in: ask
> |   10 people, see what the majority will opt for).
> 
> Better than Word, maybe.  Better than _any_ document processing system?  Not
> necessarily!

Of course not.  But very likely better than any WYSIAYG-system.  Doing
the markup with LaTeX means the system actually knows what the
different entitys stand for, which it usually doesn't with WYSIAYG.

> [...]
> |> Add to that the coming of Unicode, which effectively solves (in theory
> :-)
> |> quite a few of the problems formatting languages grappled with in
> |> the past,
> |
> |I think we should not confuse formatting (what should it look like)
> |with _input_-encoding.  Two different things entirely.
> 
> Not entirely.  There are certain problems which had to be solved by
> formatting languages or word processors in the past that are completely (or
> partially) obviated by Unicode.  

As in: how do you write \"a.  True.  But that's just a problem of
input-encoding, nothing to do with typesetting.

> Unicode all but solves the problems of:
> special characters and foreign scripts; combinative forms; code space for
> user-defined characters or codes.

Yeah, but that's not what typesetting is all about.  It's about types
(fonts, the layout on a printed page).  Having more boxes for more
types might make it easier to find a particular type, but ultimately
it's all about selecting that type in the right font / size / shape,
and to position it correctly.

> |> plus the fact that the reasons for having a plain-text source file format
> |> have now pretty well gone away, and it all makes traditional formatting
> |> languages look a bit obsolete.
> |
> |Hardly.  Using Emacs I can manipulate text in ways simply not possible
> |with a word-processor (one of the simplest examples being that in this
> |sentence I wrote word-processor only once, and wrote "Beispiel"
> |instead of example.  Emacs wrote the second and third word-processor
> |after I typed wo<shift><TAB>, and translated the German word Beispiel
> |into it's English equivalent.  Erik Naggum once reported that he got
> |functions to turn first person sentences into third person, or
> |statements into questions and vice versa.)  Emacs also reminded me to
> |close the parenthesis (bracket?) above...
> 
> 
> Ridiculous!  How are any of these things not possible in a word processor?

Well, what about: currently not implemented?

> Indeed, Word actually does most of these things.

Translating words?  Expanding words from a few letters or a
user-defined abbreviation?  Changing the structure of sentences?  Not
my word, it doesn't.

> What I was getting at was: in the past (thirty years ago?), special (rather
> than plain-text) editors would rarely be a practical proposition, because
> computers were time-shared, had little memory (and storage), and were slow;
> communications very often assumed only 7-bit ASCII text was being sent.  By
> comparison, nowadays nearly all computers are desktops, have oodles of
> memory and storage, and are extremely fast, and have a GUI; nearly all
> communications today is binary.  Thus, special editors (e.g. semi-WYSIWYG)
> which use a non-ASCII character set are a much more practical proposition.

Oh, any decent editor can use non ASCII charactersets, and could do so
for years.  How else would my chinese colleagues send and receive
their emails?

But it's still text-based, even if it isn't ASCII.

> |And of course I can easily create (La)TeX from a program, so if I'm
> |making 100 tests on something all I need to do is write a batch-file
> |(see!) and go home, and the next morning I've got I typeset tabular
> |plus graph of my measurements (using gnuplot).
> 
> With the high-level encoding I am suggesting, it would be easier still to do
> the same thing.  Codes can be written by a program just as easily as visible
> text.  

No.
Every Unix-tool knows how to handle plain text.  Shell-scripts, awk,
sed, perl, everyone.  You can easily edit it in a text-editor if
something didn't go exactly right.

With your setup, unless every single program translates from binary to
text and back to binary, all you'll ever see is gibberish.  If you do
the translation, however (so that you are still able to use awk), then
you could just as well have used plain-text to begin with.

> My encoding would be high-level, so this technique would generally be
> a whole load simpler than plain TeX, and simpler, even, than LaTeX.

Why?

> |> However, there is, now more than ever, a need for a truly standard
> |> 'enhanced' document format, that would provide for the standard encoding of
> |> a document's 'logical' structure (paragraphs, headings, list items,
> |> (floating) table rows and columns, etc.).
> |
> |This sound pretty much like a description of LaTeX,
> 
> Yes, the idea is something similar to LaTeX, but slicker, and slightly
> higher-level.  The trade-off would be that you would need a special,
> hand-holding, editor to edit these files, but the encoding would be more
> efficient (most commands carried out by a single 16-bit code), 

Sounds like it's going to conflict with the use of unicode.

> |which somehow
> |makes me doubt that:
> |
> |> (The fine details of the
> |> implementation of that structure would, of necessity, be non-standard.)
> |
> |After all, LaTeX is pretty standard indeed...
> 
> 
> My suggested encoding would be higher-level than LaTeX.  This would have the
> benefit of simplifying the encoding, at the cost of de-standardising the
> details of the implementation (e.g. fonts used, paragraph indentation and
> spacing, numbering style, etc.).  The user would certainly be able to
> specify these things, it's just that the way in which they would be
> specified would be non-standard.  Again, this is a trade-off.
> 
> [...]
> |> and redefine a lot of presently useless or
> |> ambiguous control characters (BS, HT, LF, VT, FF, CR, ESC, DEL, and maybe
> |> others) for useful purposes.
> |
> |Why?  It's not as if they really got any place in a decent text-file
> |anyway...
> 
> 
> An odd statement (or an apparently irrational one, at least).  These
> character codes (and plenty of others, thinking about it) could be re-used
> for useful and well-defined purposes, instead of being the total dead weight
> most of them are now.

Well, they _are_ controll characters.  You would need them in order to
connect your terminal to your computer (use your modem, whatever), so
you can't really get rid of them without a replacement.

> |Not that I know what all this has to do with litprog, never mind Ada...
> 
> 
> It has to do with document processing, which is important to literate
> programming, and literate programming is potentially of significance to Ada
> (and most programming languages).
> 
> |Sven, who rather likes LaTeX and plain ASCII.

> *Sven prefers the WYSIAYG alternative, as in "what you see is all you get",
> a Kernighanism I believe.  This was to do with a different argument (about
> WYSIWYG word processors not encoding information about the logical structure
> of a document.  Of course, nowadays, they all do, or can do, this.)

No doubt they could, if you entered it beforehand.  Which somehow
defies the idea of WYSIAYG.  And since the above is now out of
context, I would like to clarify:  I _hate_ WYSIAYG (or WYSIWYG).

Well, back to reinstalling Windows...

Sven

(*) Urban legend alert!
-- 
 _       _   Lehrstuhl fuer Mustererkennung und Bildverarbeitung
| |_ __ | |__                                                        Sven Utcke
| | '  \| '_ \   phone:      +49 761 203 8274                   Am Flughafen 17
|_|_|_|_|_.__/   fax  :      +49 761 203 8262           79110 Freiburg i. Brsg.
mailto:utcke@informatik.uni-freiburg.de   www.informatik.uni-freiburg.de/~utcke