RE: Document Engineering was RE: Way OT: Adam Smith and SoftwareMarkets

["Robert C. Leif" <rleif@rleif.com>]

DTDs can not specify ranges. As I have stated(1), one can fake Ada with XML
schemas. Admittedly, one can also create unreadable code. I create my
schemas by importing multiple schemas, which all have their own prefixes.
I have two papers in press describing this approach to the creation of
analytical cytology for medical informatics.
(1) R. C. Leif, SIGAda 2001 Workshop, "Creating a Symbiotic Relationship
Between XML and Ada" Ada Letters Vol. XXII, pp 24-41 (2002). 
Bob Leif
-----Original Message-----
From: Georg Bauhaus [mailto:bauhaus@futureapps.de] 
Sent: Thursday, March 06, 2003 2:41 PM
To: comp.lang.ada@ada.eu.org
Subject: Re: Document Engineering was RE: Way OT: Adam Smith and
SoftwareMarkets

Robert C. Leif <rleif@rleif.com> wrote:
: Another way of stating this is Document Engineering should be a branch of
: Software Engineering. I might note that a common editor for text and
source
: code with bidirectional hyperlinks would be of use for software
development.
: This capability is included in the XML specifications.

A related issue that I think might be interesting for Ada programmers is
the relationship between XML Schema datatypes and Ada's types and
constraints. You can have the rough equivalent of

 Generic_Bounded_length(max => 32)'s Bounded_string type
 subtype Min_to_max is Integer range 100..100000;

and more, an overview is at
http://www.xml.com/pub/a/2000/11/29/schemas/dataref.html

But, when does it help? Say you wanted a constrained datatype
for paragraph width, that can be easily turned into an Ada
subtype. However, a paragraph is always inside a frame, which
in turn has a width that may vary. What is the maxExclusive
restriction for paragraph width? You can't really give one
without referring to the frame's maxExclusive. How do you
do that? Does it make sense, both inside the Schema and at
the interface between the Schema and your Ada application?
After all, constraints in Ada can be dynamic.

Likewise you can specify patterns that values (XML data) will have
to match.  But is this enough? Is pattern machting enough or do
we need parsing anyway, for 'Valid input? Examples include the
xs:date restriction, which might not be enough to match the real
world date conventions--unless you can force the data producers
to write dates in the required format...

So it might be better to think twice whether the situation is
such that Schemata will solve the data format issues best, or
whether plain old DTDs provide much needed flexibility. Perhaps the
ATTLIST can be supplemented with data representations information
when needed.


-- Georg

----------------------------------------------------------------
Georg Bauhaus                              bauhaus@futureapps.de
Future Apps GmbH
phone: +49 203 306 1560