Sunday, October 28, 2007

Using Requirements Management Tools in Software Product Line Engineering: The State of the Practice

Last week we discussed the paper "Using Requirements Management Tools in Software Product Line Engineering: The State of the Practice" which was published in SPLC 2007. The paper analysis the current scenario of requirements tools being used in Germany companies, and identifies the tools weakness to support software product lines requirements. As result, the authors proposes a set of requirements for requirements management tools.
The majority of the authors work in the industry, and because of that they did not define a systematic approach to do the analysis. They said that it was all based on practical experience, but should it be enough? Doing it does not increase the chance of bias in the research?
Besides, the requirements defined for requirements managements tools were too superficial described, some lacked of reasons why to include it, while others did not explained how it could aid the a software product line process.
On the other hand, the paper was derivated from a report, which may explain what was not very clear in the paper.


Ricardo Cavalcanti said...

they analized the three tools in the market, but did no questioning whether would it be better to adapt one of those tools, or build a new one for the purpose of SPL. I'd bet you at RiSE found it better to build a new one! =)
Still, haven't you started ToolDAy's develpment, would it be feasible to extend existing tools like caliber? Is there any kind of common format in which different tools can talk together?

Eduardo Almeida said...

About the paper, I agree with Lica. I would like to see a more systematic evaluation mainly with research papers. I think that the industrial tools is part of the analysis. But, for example, and about the criterias to choose them? On the other hand, I liked of some references cited on the paper, especially, the survey involving 100 requirements.

Lica said...

Ricardo, I've never seen the tools the paper mentions, except the RequiLine. But I know that is possible to have work together with tools, like CalibreRM and Doors, because the Pure::Variants has it.

Vinicius Garcia said...

I agree with Lica and Eduardo's opinion. I think that we need more systematic evaluation in order to make our research more practical. The criteria used by the authors to identify and select the tools aren't clear for me.
The possibility to work with others SPL tools is a great feature that we can investigate to develop it (in a new version of ToolDAy)

Danuza Neiva said...

The paper suggestions have high level of abstraction in some points. For example, it don't detail how should be the traceability, also it don't handle the traceability to variability model and views, nor the consistency of the information between differents views.