Lists Home |
Date Index |
No size fits all. Sure.
And you are right, lots of systems, most in fact,
worked without the ESIS. But the interoperability
of the systems was poor in the extreme with the
exception of the operation of exchanging SGML
and even then, one needed expert knowledge to
achieve that seamlessly.
Ok, but it meant buying an SGML system that was
production worthy was typically a one-system purchase
and it was expensive. SoftQuad tried to break that
with low cost editors, but went broke doing that.
Two things about interchangeable parts systems:
1. Costs usually come down over time if the
manufacturing volume is reasonably high.
2. Systems can be tuned within reasonable
limits to missions without too much custom
coding by mixing an matching parts.
I don't dispute the value of custom coding;
just saying it isn't needed for the majority
of the work we do. As I said to David, the
world of SGML was big systems and specialists
using publishing systems and not networks.
That is not the world of XML for the most part.
As a result, if XML is to be a core technology,
it has to be very resilient and reliable in the
face of not-quite-ADEPT hands.
From: Gavin Thomas Nicol [mailto:firstname.lastname@example.org]
On Wednesday 26 February 2003 04:00 pm, Bullard, Claude L (Len) wrote:
> Walter, as much as I respect you, this is like
> saying all rifles should be handmade: in other
> words, no interchangeable parts.
The point is that even though *some* uses of rifles benefit from mass
production, it is unreasonable to expect *all* rifle components to be
interchangeable. After all, a Lee Enfield is great as a bush rifle, but I
wouldn't want to use it if I was a sniper.
> The days of SGML-pre-ESIS were a nightmare because of that.
Sure, but you also know that even though a lot of good things were done with
the ESIS, many/most *applications* didn't use it.