RE: Data versioning strategy: address semantic, relatio nship,

But that is the point:  is the question of process where a version has to know about the impact or the observer has to know?   That is the question I am posing to Roger et al.   Is versioning observer or user or process or test (all observers really) independent?

A version obviously impacts the system where it is installed, but a version is not cognizant of the change impact.   What follows?  Version change control is not the same for all observers (same input, different controls, different outputs).  What you would not want is to have multiple controls over the same process.  That is the classic recipe for chaos.

The reason for this part of the thread is to examine the strategy as realized in the tools.  Practically, that is the answer.   This becomes problematic when a production system tool (such as a RAD) is combined with a version control tool (say CVS vs SourceSave) that have different critical variables or emphases.   The production team has to understand the fine points and adjust (eg, fine grained vs coarse grained commits and updates).

len

-----Original Message-----
From: Greg Hunt [mailto:greg@f...]
Sent: Tuesday, December 11, 2007 1:25 PM
To: Len Bullard
Cc: Costello, Roger L.; xml-dev@l...
Subject: Re: Data versioning strategy: address semantic, relatio nship, and syntactic changes?

Actually Len, I was observing that if you use phase changes as an analogy for change control, phase changes have to be dependent on the observer, because different observers experience the impact of changes in radically different ways.  I was not suggesting that that happened in the natural world (hence the reference to Heisenberg in the original email).  The majority of change problems are not in getting out of test, they are in assessing the impact on the recipient.