Section: New Results

Interactive Code Restructuring

Participants : Cédric Bastoul, Oleksandr Zinenko, Stéphane Huot.

This work falls within the exploration and development of semi-automatic programs optimization techniques. It consists in designing and evaluating new visualization and interaction techniques for code restructuring, by defining and taking advantage of visual representations of the underlying mathematical model. The main goal is to assist programmers during program optimization tasks in a safe and efficient way, even if they neither have expertise into code restructuring nor knowledge of the underlying theories. This project is an important step for the efficient use and wider acceptance of semi-automatic optimization techniques, which are still tedious to use and incomprehensible for most programmers. More generally, this research is also investigating new presentation and manipulation techniques for code, algorithms and programs, which could lead to many practical applications: collaboration, tracking and verification of changes, visual search in large amount of code, teaching, etc.

This is a rather new research direction which strengthens CAMUS's static parallelization and optimization issue. It is a joint work with two Inria teams specialized in interaction: EX-SITU at Inria Saclay (contact: Oleksandr Zinenko) and MJOLNIR at Inria Lille (contact: Stéphane Huot).

In 2015, we presented our interactive tool, Clint, that maps direct manipulation of the visual representation to polyhedral program transformations with real-time semantics preservation feedback. We conducted two user studies showing that Clint's visualization can be accurately understood by both experts and non-expert programmers, and that the parallelism can be extracted better from Clint's representation than from the source code in many cases [21] . We are planing a first release of that tool in the coming year.