LATEX-L Archives

Mailing list for the LaTeX3 project

LATEX-L@LISTSERV.UNI-HEIDELBERG.DE

Options: Use Forum View

Use Proportional Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"William F. Hammond" <[log in to unmask]>
Reply To:
Mailing list for the LaTeX3 project <[log in to unmask]>
Date:
Fri, 27 Nov 1998 10:14:22 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (111 lines)
Sebastian Rahtz <[log in to unmask]> writes:

: William F. Hammond writes:
:  > This could as well be:
:  >
:  > \documenttype{article}
: which is not valid XML, so whats the point?

But haven't I been saying here for several weeks that I think it more
sensible to use the additional strength of sgml relative to xml while
working on an authoring platform?  It is, upon transliteration -- but
certainly not char-by-char transliteration -- valid sgml under the
didactic gellmu dtd, which is subordinate to a non-reference sgml
declaration.

(Because I've been using different declarations I can become confused
about what is legal when, which is why I made the slip about case
sensitivity that Sebastian kindly pointed out several days ago.)

And didn't I say that it could trivially be processed to xml?

And we understand, don't we, when xml made from my sgml dialect is
*parsed*, the parse stream looks just like the parse stream from the
sgml.  (Well, one needs to do something about "eqnarray*" if that name
is used.)

:  > If I run this markup through my elisp, I get the following sgml:
: but i want the *source* be to useable, not a result munged by  some
: script!!!

We do understand, don't we, that elisp under GNU Emacs is not just
a scripting language but rather a full lisp that can be run either
interactively in Emacs or else in batch mode.  We're NOT talking about
clever merging of "sh", "sed" and "gawk".  And we have, moreover, the
possibility of layered error diagnosis with the elisp finding low
level syntax mistakes and the parser finding structural mistakes.
Then, if, for example, validation fails for an output sgml or xml,
it's most likely a problem with the sgml processor for that output.
Or if LaTeX yells, it's most likely a problem with the sgml processor
for that target.  (I do, however, as I write new things, still find a
few authoring mistakes at that stage.)

Wasn't it previously agreed that good LaTeX is just inches away from
something usable?  If so, then I expect that something usable when
coded for *personal* convenience to look like LaTeX is also usable.

Of course, if one is happy writing verbose xml, then one does that.
It's just that since I have this persistent LaTeX habit and find this
a convenient way to write sgml that can be robustly translated to xml,
I thought that others might also find this to be a personal
convenience.

(Of course, if I were designing a wysiwyg interface, then I expect that
I would go for the xml.)

Finally, there are four things for authors:
 (1)  editing format.
 (2)  exchange format.
 (3)  private archiving format.
 (4)  public archiving format.

There is also the issue for any format that is a printable plain text
format whether it is sane for *backup to paper* with recovery by
optical scanning.

These formats do not all have the same purpose.  What is optimal
for one purpose may not be optimal for another.

The question then always is: starting at X can I safely get to Y?

: but it goes through you (inevitably flaky) elisp. thats the flaw.
:
: if you *started* with valid *ML markup, wouldnt it be even easier?

But I *am* starting with valid sgml markup.  That's why it is called
gellmu and not called LaTeX.  It only looks like LaTeX.  I must have
my DTD in mind while writing.

My elisp may be flaky if applied to arbitary LaTeX documents.  I did
not design it primarily for legacy documents but rather for being able
to use LaTeX-like markup to create sgml documents.

For example, inasmuch as "frac12" is a legal name in sgml, my elisp
regards it as a legal name.  If I want to use something obfuscated
like "\frac12" as a synonym for "\frac{1}{2}", then I must have that
name in my DTD and my processors must know about it.  (I choose not to
go there, but one could.)

I did say that my choice to use LaTeX-like markup to create sgml
is a personal matter.   On the other hand, it staves off my symptoms
of "withdrawal" from editing real LaTeX (which I still very much
shall continue to use to get to paper).

And when I come across a good legacy document such as the LaTeX3
prospectus (part of the LaTeX2E distribution) by Mittelbach and
Rowley, it is not that much work to make it legal sgml via the
transliterator.  From that sgml an equivalent latex document may
be re-constructed.  Or a good html version may be made.

If I am editing for the transliterator, I do need to be minimally
aware that I am writing sgml.

:  > Note that this approach is different from that of James Clark's "jade"
:  > which "centralizes" style for all "backends" using a DSSSL stylesheet
: please don't blame James Clark in person for the fundamental concepts
: of DSSSL, an ISO standard developed after a decade of work...

Blame?  There was no blame from me.  I am a fan.

                                   -- Bill

ATOM RSS1 RSS2