Study of defects in a program code in Python

In the paper, defects in a program code in Python are considered. It is shown that these defects are different from those in a code in C/C++; hence, there is a need in study of defects in large-scale projects with an open source code. A classification of the defects found, which is based on whether type inference is required for finding an error, is presented. It is shown that there exists a small portion of “simple” defects; however, the determination of the majority of the defects requires type inference. The question of what constructs of the Python language are to be supported in type inference for finding real defects is discussed.