Re: [xml-dev] Text Markup Part II

[Peter Flynn <peter@silmaril.ie>]

On 20/01/12 12:02, David Lee wrote:
[...]
> * Was there a slow transition of adding markup then some kind of "Ah Ha!"
> moment where people realized they needed/should wrap the whole thing in a
> tag ?

I don't think there was a specific road-to-Damascus moment, but the 
development of Script--GML--SGML and the start-tag...end-tag very 
roughly paralleled the development of TeX (grouping) and LaTeX with its 
\begin...\end markers. But even roff/Runoff had a rudimentary 
turn-on...turn-off mechanism. Wrapping the whole thing is largely an 
[S]GML and LaTeX thang:

<!doctype foo system "bar.dtd">
<foo>...</foo>

and

\documentstyle{bar}
\begin{document}...\end{document}

Even COCOA markup assumes the whole document is being treated the same way.

> * Was there a legacy/historic (SGML?) reason why documents needed to be
> tagged from the very beginning in order to parse tags at all ? (A kind of
> inline file header/type/magic number ?)

Yes: if you want to know what's coming up, you need (a) a roadmap (DTD), 
and some signposts (markup). Without that, you might as well point the 
car at the sun and drive until you find a nice place for a picnic. 
Exploration by serendipity is fun, but it ain't going to get you a 
consistent document, and don't forget most of these developments were 
driven by the need to print and publish, where consistency is a requirement.

> * Was there a natural evolution from tagging little bits of text to the
> point where it was felt that all of it should be tagged ?  Or perhaps there
> were cross-purposes from different camps ?

I don't think there was an evolution (natural or unnatural) from 
fragments to whole documents. If you're going to do some of it, you may 
as well do all of it and be consistent. All the early writers on markup 
make the assumption that the document is being treated as a whole.

> What I'm trying to get at is the fundamental rational between  what appears
> to be two extremes not necessarily compatible, and why we ended up with only
> the later.
>
> A) "Marking up bits of text where needed"
> And
> B) "A full text markup vocabulary for encoding an entire document"

I'm finding it hard to see a use case for (A) at all. I'm not aware of 
it ever being done.

> My 'gut feeling' on this, not having been privy to the entire history ... is
> that it was an artifact of processing.
> That (it was felt that) processors simply needed to detect the *possibility
> of markup* within text at the start of a document

I'm not sure that has ever been the case, but I would be extremely 
interested if you have an instance of it.

> and thats how we ended up
> wrapping the whole thing in a root element , thus crossing the threshold
> from "A document with marked up text" to "A Markup formatted/schema based
> document for encoding text".

Was there ever any such threshhold? I was just a student in the days of 
roff/Runoff, and the only distinction given to those files was that the 
filename ended in an extension that signified the processing expected 
(eg .rno), so in a sense the presence of markup in the text was 
unheralded per se in the text itself, except that you tended to start 
with all the settings, so /etc/magic would tell you that ^\.[a-z] meant 
it was a Runoff document, the same way that ^\\doc told you it was LaTeX 
and ^\<\! told you it was SGML.

///Peter