Re: XML and Lout already work well

[Greg A. Woods]

[ On Wednesday, September 19, 2001 at 17:52:04 (+0100), address@hidden wrote: ]
> Subject: XML and Lout already work well
>
> Everything looking the same is what communication is about. Why do 
> you use the roman alphabet when you could create your own letter 
> shapes?

I think others have answered this question much better than my earlier
attempt.

Clearly having all the world learn to speak English as a first language
would greately facilitate communications.  However one cannot even begin
to estimate the loss of culture this would entail.

In computing the problems are even more critical -- loss of culture
equates to loss of knowledge of the many and varied techniques at
problem solving.  If all the world wrote all their documents in XML
using the same DTD, where do you think you'd be?  If all the world were
to have stayed with FORTRAN as the only programming language where would
computing be today?  If all the world were forced to switch to C++ would
we ever get anywhere?  (Mind you I wouldn't cry if all the world were
forced to switch to Smalltalk!  ;-)

There is a place in the computing world for new ideas and new languages.
Unlike in human languages there's still infinitely more to explore in
the variances in computer-human interaction.


> A disadvantage of Lout is that it allows macros which allow 
> sloppy markup. Eg it's common in Lout to do:
> 
> @PP A quick brown fox      (1)
> jumps over the lazy
> fox.
> 
> Where the role of @PP is confusing.
> 
> Of course, Lout could also allow:
> 
> @P {A quick brown fox      (2)
> jumps over the lazy 
> fox.}

Good point.  But that's about as far as you have to go in this
discussion w.r.t. lout.  Lout allows you to to be pedantic about your
syntax if you want, so what's the problem?

> XML requires you to do it the nice way, so if you like formal 
> things XML is very slightly better.

So forget lout and use XML exclusively.  Nobody forces you to use lout
or to learn its syntax!  ;-)

> BTW, I've always been puzzled 
> by Jeff's assertions in the doc that form (1) is better than (2).

To the average user the meaning of the stand-alone address@hidden' is clear and
there's no real reason to use the more complex (and thus more error
prone) form.

> To me (2) is much clearer and pleasant than irrational (1) which 
> breaks the nice functional nature of the markup, but then 

#1 isn't "irrational" -- it's quite meaningful from the document's point
of view:  start a new paragraph.  Why use the more complex and more
error prone form when the simple one suffices?

> I'm the kind of person to always end my HTML paragraphs with </p>
> even when they're not required. Is it just me who thinks (2) is 
> better than (1)?

Me too, though in modern HTML that practice is explicitly deprecated.

> Uniform syntax allows you to use tools that understand that uniform 
> syntax. Many tools whose semantics is language independent can be 
> factorised thanks to a uniform syntax.

Indeed.  So the task is to write an XML to lout compiler then, no?

Changing lout to use XML syntax is, quite simply, WRONG.  Adding XML to
lout in parallel with the existing lout language would be pointless as
grafting a giraffe and elephant together.  Who'd want big floppy ears
and a trunk at the top of such an excessively long neck?

> If you'd design a new typesetting system today, you'd probably 
> use XML at least for some markup

I doubt I would.  As a programmer I really do really really really HATE
all forms of SGML-derrived markup syntax.  From a programmer's point of
view it's littered with dozens of stupid unnecessary ugly lexical
elements!  It's far worse than PostScript IMNSHO.  It's just too
inelegant to use directly from the keyboard and all the redundant crud
is really useless and difficult to deal with internally in any data
handling system.  There are FAR better ways to canonically represent
data in a flexible and syntactically pleasing way.

> but retrofitting it into Lout 
> does not seem worth the effort, and having two syntaxes 
> contributes to code bloat and makes it harder to understand 
> each other. I think that having two syntaxes built-in is worse 
> than one non-standard syntax. Of course preprocessors are fine,
> if of little use.

Why do you say a preprocessor (I think it would have to be more properly
called a compiler though) is "of little use"?

If what you say about the benefits of using common XML syntax in your
domain is really true then isn't a XML->lout compiler a useful thing?
It allows you to translate your XML documents into well formed
PostScript for printing!

-- 
                                                        Greg A. Woods

+1 416 218-0098      VE3TCP      <address@hidden>     <address@hidden>
Planix, Inc. <address@hidden>;   Secrets of the Weird <address@hidden>