The objective of the ERC PoC SAFE (ga. no. 966733) was to significantly broaden the applicability of a software solution developed in the course the ERC CoG SYMBIOSYS (ga. no. 647295) to wider ranges of software and assess the commercial viability of such a technology. The solution developed as part of the SYMBIOSYS project is a procedure that can automatically test multi-threaded software and identify concurrency bugs, a class of hard-to-find programming errors. Multi-threaded software are programs written to utilize multiple threads of execution, that is instructing the processor that runs it to parallelize computation by explicitly splitting up the work to be performed across multiple parts that may be executed in parallel. Concurrency bugs (e.g. data races, deadlocks, etc.) are bugs that manifest in parts of a faulty program that are executed in parallel and largely independent of one another. To avoid them, programmers have to carefully consider the interaction between all independent parts and add synchronization wherever necessary to prevent inconsistencies between them.
For the technical part, the C and C++ programming language ecosystem was surveyed. The key finding was that apart from a small core, there is no set of concurrency frameworks or libraries that dominate the industry, although many high-level concepts are widely applied. To extend the applicability of the prototype implementation on a wider range of software, the engineering was focused on more low-level aspects that had not been considered by previous research efforts. Development also produced an auxiliary tool that eases application of new test targets, as well as an optimization of deterministic memory allocation, which is of particular importance for the approach.
In terms of commercial viability, a market analysis was conducted, offering insights into competitors and the overall target market. Several business models were outlined and considered. While there exists a yet unmet need for a technology similar to the one offered by the SYMBIOSYS solution, there are ultimately more challenges that need to be addressed before a marketable product can be defined and developed. This is in part due to research questions that came up during the project, but also to a large extend due to the unique challenges posed to the technology by the high fragmentation of the targeted ecosystem.
