Abstract
The advantages a DSL and the benefits its use potentially brings
imply that informed decisions on the design of a domain specific language are
of paramount importance for its use. We believe that the foundations of such
decisions should be informed by analysis of data empirically collected from
systems to highlight salient features that should then form the basis of a DSL.
To support this theory, we describe an empirical study of a large OSS called
Barcode, written in C, and from which we collected two well-known ‘slice’
based metrics. We analyzed multiple versions of the system and sliced its
functions in three separate ways (i.e., input, output and global variables). The
purpose of the study was to try and identify sensitivities and traits in those
metrics that might inform features of a potential slice-based DSL. Results
indicated that cohesion was adversely affected through the use of global
variables and that appreciation of the role of function inputs and outputs can be
revealed through slicing. The study presented is motivated primarily by the
problems with current tools and interfaces experienced directly by the authors
in extracting slicing data and the need to promote the benefits that analysis of
slice data and slicing in general can offer.
imply that informed decisions on the design of a domain specific language are
of paramount importance for its use. We believe that the foundations of such
decisions should be informed by analysis of data empirically collected from
systems to highlight salient features that should then form the basis of a DSL.
To support this theory, we describe an empirical study of a large OSS called
Barcode, written in C, and from which we collected two well-known ‘slice’
based metrics. We analyzed multiple versions of the system and sliced its
functions in three separate ways (i.e., input, output and global variables). The
purpose of the study was to try and identify sensitivities and traits in those
metrics that might inform features of a potential slice-based DSL. Results
indicated that cohesion was adversely affected through the use of global
variables and that appreciation of the role of function inputs and outputs can be
revealed through slicing. The study presented is motivated primarily by the
problems with current tools and interfaces experienced directly by the authors
in extracting slicing data and the need to promote the benefits that analysis of
slice data and slicing in general can offer.
Original language | English |
---|---|
Title of host publication | Proceedings of Workshop on Knowledge Industry Survival Strategy Initiative |
Publication status | Published - 2009 |