Seminar

We present experience in modernizing the software infrastructure of the Baikal-GVD neutrino telescope in two directions: reengineering of legacy shower reconstruction code and evolution of the data processing automation system. The first direction is the migration of Fortran-based charged particle shower reconstruction into a modular C++ task pipeline built on the BARS framework. Key solutions include using common blocks as a ready-made data specification for container design and incremental replacement with verification at each step. Incremental replacement preserved functional suitability, while structuring the stages as independent tasks improved maintainability and usability. The second direction is the evolution of the data processing automation system. The existing architecture, while preserving its functionality, was extended: task dependencies were separated from source code into configuration files, allowing processing graphs to be modified without changing task code. New capabilities — integration with distributed storage (EOS) and batch processing system (HTCondor) — were implemented through the Law extension, which does not affect the working code. These solutions improved modularity, reusability, and error protection. The resulting system provides batch processing of large volumes of accumulated data on a computing cluster.

The achieved results are analyzed against the quality characteristics of GOST R ISO/IEC 25010-2015. Principles of maintainable scientific software are formulated, applicable in related areas of experimental physics.

Information on the seminar is available at Indico.
Сonnecting to MTS Link.