Fujitsu: Technology of change tracking in software open source

In May, 2016 Fujitsu announced development of technology which quickly finds the changed fragments of the source code, including again added functions, for acceleration of project implementation using open source software.

The Open Source Software (OSS) has free access to the source code which can be repeatedly changed with adding and removal of functions without exact fixing of the processed code locations. OPO is widely applied in projects of software development if it is necessary to prepare very quickly a system under requirements of customers. When developing using OPO the part of changes can be made by certain users, however there is also a set of changes which become in parallel a large number of other developers. It means that the user should confirm corrections in the source code before changes of the software. So changes will need to be checked in tens of thousands of files of the source code regarding detection of text distinctions or by visual check of differences in structure of calls of functions that will demand considerable time and efforts.

At detailed verifications of the source code manually, for example, after detection of distinctions, it is possible to enter a part of function name and to narrow areas for check, using function of selective search which displays the options corresponding to this fragment of the code for each separate case. In large-scale OPO there is a huge number of functions which need to be checked, and there is a set of functions with similar names, in particular, method names in classes, and such search takes too much time.

Fujitsu developed two key technologies for effective change tracking when updating large-scale software with the dug-out code.

1. Technology of operational comparison of ways of a call of functions with graphic display

At change tracking of the source code between different versions of the same software the attention addresses on the way of a call of a certain function. Based on this technique, Fujitsu developed technology which, by comparison of differences between two versions of the source code in each point on the way of function call, specifies the purposes for comparison and displays them graphically for a clear understanding. Such approach allowed to reduce by 44% time necessary for selection of distinctions, in comparison with the existing techniques of the joint comparison.

2. Technology of effective search of similar names

In large-scale OPO there is a set of functions with identical names. At input for search of function name there was an opportunity to perform search of coincidence for those parts which have the high level of reaction. It became possible due to development of a data structure which numbers and in advance keeps the corresponding search results for different combinations of the entered characters.

Thanks to these developments became possible systematically to execute search of the changed points OPO on the basis of appropriate ways of a call of functions. As a result, for the program containing 3.8 million code lines and about 400 changed points, time spent on search of points was reduced by 4 times in comparison with the previous attempts of search: from 26 to about 6.5 o'clock. Fujitsu hopes that it will help not only to reduce expenses on development, but also to popularize use of software open source. In the future considerably the quantity of supported languages, including scripting languages will extend, and commercial distribution of technology of Fujitsu is going to begin in 2016 financial year.