Visualizing Objects and Memory Usage

Mariano Martinez Peck (1, 2) , Noury Bouraqadi (2) , Marcus Denker (1) , Stéphane Ducasse (1) , Luc Fabresse (2)

1 RMOD - Analyses and Languages Constructs for Object-Oriented Application Evolution
2 CERI SN - IMT Nord Europe - Centre for Digital Systems

Most of the current garbage collector implementations work by reachability. This means they only take care of the objects that nobody else points to. As a consequence, there are objects which are not really used but are not garbage collected because they are still referenced. Such unused but reachable objects create memory leaks. This is a problem because applications use much more memory than what is actually needed. In addition, they may get slower and crash. It is important to understand which parts of the system are instantiated but also which are used or unused. There is a plethora of work on runtime information or class instantiation visualizations but none of them show whether instances are actually used. Such information is important to identify memory leaks. In this paper, we present some visualizations that show used/unused objects in object-oriented applications. For this, we use Distribution Map which is a visualization showing spread and focus of properties across systems. We extend Distribution Maps to represent the way classes are used or not, since we distinguish between a class that just has instances from one that has used instances. To identify unused objects, we modified the Pharo Virtual Machine.

Domaines: Langage de programmation [cs.PL]

Mariano Martinez Peck, Noury Bouraqadi, Marcus Denker, Stéphane Ducasse, Luc Fabresse. Visualizing Objects and Memory Usage. Smalltalks'2010, Nov 2010, Buenos Ares, Argentina. ⟨inria-00531510⟩