Friday, January 22, 2010

Of all models a W-model

Last Wednesday I attended a meeting from the Dutch Testing Organization TestNet related to W-models. During this meeting 3 guest speakers: tried to sell their vision about a "W-model".
Of course we all are aware of the V-model. We are also aware of reviewing and how it can contribute gain faster a vision about the quality.

Egbert Bouman: explained his view towards a SmarTEST W-model which represents more the iterative approach of development and the interaction with business
Jan Jaap Cannegieter: informed us about the historical mentioned W-model referring to Paul Herzlich who already spoke about a W-model in 1993 (source: http://gerrardconsulting.com/?q=node/531). He explained why and how reviewing methods adding to the well know V-model created a "W". He is also one of the co-author of a very good book about reviews (Dutch: "Reviews in de praktijk") In his W-model the moments of review results in the "W-model"
Jos van Rooyen: Who referred to his experience from already 15 years ago referring to a similar model where he added the products to be reviewed and/or the roles to perform to a similar W-model and created a "triple V-model"

The outcome of the meeting was to choose for one of the models to be the W-model. Unfortunately history was overruled by reality? The model of Egbert was chosen.

Based on that event I cam up with some questions:

1. I always thought about the V-model as a development model. I might be a misperception from me. In the Dutch testing world we refer to the V-model too often to define and explain the testing process. We even spend time to proof that the V-model fits in other development models. Can this be done to make a development model fit into another? And why also spent this energy?

2. Another thing I wonder is why are we referring to a W-model if we speak about checking the test documentation and other information sources?

3. Do we need a model to refer to the importance of checking documentation?

Some Assumptions
Assume a V-model is a development model like DSDM, RUP, SCRUM, Spiral. If the W-model is created by adding check moments to a V-model. The this W-model cannot always be used as we don't use the V-model always. This means that addding check moments can also be done to those other models. This will results into other presentations of those models too. When doing this there is no specific need to refer to a W-model as in other models you cannot draw the "W".
Assume the V-model is not a development model. It is a representation of a model like de sequential Waterfall model. Then again, if a processes of the sequential model is lifted to create a "V" we can always put them back into position so we got 2 parallel lines. In this case, the "W-model is only usable for other development models which can be supported with the V-model. Basically this can be done. Only it would be hard. Is in this case also worth to spend energy to translate to a W-model.



I would suggest to refer to the W-model to explain how reviews/checks can be embedded in a model. Only not refering to a W-model as model for full implementation.

Some other resources:

W-model from Andreas Spillner: http://squac.iti.upv.es/JTS/JTS2004/docs/Wmodel.pdf
Similar posting of W-models: http://www.testingthefuture.net/2009/09/the-w-model
The Dangerous and Seductive V Model: Nice twist here is the reference towards Morton's butterffy-model: Morton, S. (2001). "The Butterfly Model for Test Development". Sticky Minds website. Accessed 6th October 2008

3 comments:

  1. Nice wrap-up and a couple of good questions Jeroen! We should do some further talking. Bye, Egbert

    ReplyDelete
  2. gr8 stuff
    thanks for sharing your knowledge
    http://soft-engineering.blogspot.com

    ReplyDelete
  3. Thanks Egbert for your comments,
    perhaps on a future testnet meeting or sooner it would be interesting to discuss further on this topic.
    regards,
    Jeroen

    ReplyDelete