Variant Management at the BCS CMSG Conference
Today was the final day of the 3rd BCS Configuration Management Group Conference.
My paper on Variant Management, co-written with Danilo Beuche of pure-systems, was accepted a few months ago and today was the point of delivery.
As noted previously there were a few sleepy heads around but I went easy on the audience (at least at the start).
I’ll put the paper up on the Software Acumen web site once I’m back in the office but the basic idea was that there are sub-optimal ways of managing variant development and there are optimal ways of managing variant development.
Sub-optimal methods including Clone-and-own, Independent Component Teams and Platform Versions. These can help in some cases - OK, maybe clone-and-own only ‘helps’ for a very short time - but each has their disadvantanges. One issue of course is that unlike Variant Management they don’t necessarily manage variants across the life cycle but are rather focussed on managing variation in the Configuration Management, Build and Release processes.
When presenting I prefer to have as much audience interaction as possible - it is easier for everyone - so I peppered the talk with questions to the audience and adapted the line of discussion to people’s experiences. We had a good discussion of how variant management helps address the problems of the other named approaches by allowing component dependencies to be formally captured and maintained and so helps developers minimize component interdependencies in a way that isn’t possible when dependencies are not visible or are only managed informally.
Anyway, I think the audience was largely sympathetic and I spoke to most of the people later in the day about the content.
Today was also busier in terms of people coming to the stand so this seemed like a worthwhile event - at least we were noticed and recognised by a few people - however, as always, the proof of the value of an event is in the customer contact that follows and (hopefully) an eventual implementation…
