This paper looks at what questions programmers ask during software evolution – when they need to touch the code base to undertake code change task, what do they not know and want to know? To learn this, they conducted two observational studies in a laboratory and an industrial development team – the main difference between the two being the scale of the code base. After applying grounded theory, they summarized the data to 44 FAQs which fit in four categories, namely, finding initial focus points, building on those points, understanding a subgraph and questions over groups of subgraphs. Furthermore, the paper also talks about questions in context, starting from a single question, to multiple supporting questions, to tool question and finally arriving at some results. At last, talking about the implications, the authors propose a refinement of tools for programmers that can better help them answers questions instead of offering a narrow scope only.
- Thinking about related work, e.g.[ Information Needs …], what is different with the scenario in this paper, namely code base + software evolution?
- The use of the graph model seems convenient? Where did it come from?
- Given a highly structured list of (44) questions, what can be done with that? Can we prioritize them? Can we analyze their inherent difference and design tools accordingly? etc. (And on the other hand it is really possible to use open coding to achieve such a structured result.)
- Is using tool the only way out to answer questions?