[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
Re: [xml-dev] RE: Encoding charset of HTTP Basic Authentication
- From: cbullard@hiwaay.net
- To: Peter Flynn <peter@silmaril.ie>
- Date: Thu, 02 Feb 2012 08:41:43 -0600
Contemplating the work before me, it may be advantageous to strip a
DTD down to a set of DTDs each of which do something specific (thus
the earlier thread about granularity). The problem of version creep
is it leads to a polyglot that is difficult to manage in code say in
the GUI controls where it better be a very simple set of options or
the user gets lost and codes the wrong elements.
There likely was a technology shift of focus between versions because
having coded to the most recent and now rolling back one, it is
apparent in the prior version someone(s) was trying to use database
concepts and didn't quite make it across the chasm from structured
chapters to tables and rows. In the new version, the elements in the
databasey parts of the document (reference matter) look more like
objects and property sets (someone realized not every row in a
document table really is a row in a database needing an id. Still, we
have to id every one of them (#REQUIRED) despite the fact that no
IDREF points to them or ever will. Upshot: we pay the cost (or are
paid for it) until the very slow migration to the newer version is
reasonably complete while keeping all the code supporting the last
version because.. you never know.
It will most often come back to incomplete knowledge, Boltzmann's revenge.
Possibly a different topic: how to factor the least used parts of a
language into their own boxes so those that do can and those that
don't do don't have to know how.
len
Scope Creep: often heard in the bathroom gargling before spending the
morning hassling his co-workers.
Quoting Peter Flynn <peter@silmaril.ie>:
> I have lost count of the projects whose first task was to strip all
> namespaces and cut the XSD down to a DTD representing only the
> element types that were actually used. In only a few cases have I
> been able to persuade them to do the modelling in RNG and recreate a
> minimally workable Schema. But this is solid text-document XML, with
> more character data content than markup :-)
>
> ///Peter
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]