The author successfully provides a big picture view of automatic software repair. The survey covers an impressive list of papers, giving an excellent overview of automatic software repair. Those papers are briefly reviewed in order to give readers an idea of their contributions. In general, the provided definitions are useful. Each section has a valuable explanatory introduction.

The author organizes the reviewed papers according to two main classes: behavioral repair (by modifying the source code) and state repair (by modifying the state of the running program). For each class, the bibliography is divided into several subclasses. Classifications are always (or almost always) subjective, but I think the author has done very good work. However, the inclusion of some works about software availability is questionable, that is, to keep a program available is not to repair.

Personally, I don’t appreciate the use of the term “bug”; I would have preferred “defect,” as well as IEEE and ACM software engineering definitions for error, defect, fault, and failure. I also found the use of the term “oracle” a little confusing. At the end, I was expecting some conclusions about the most promising methods, and possibly some guidance for future research (although the paper’s title does not promise this); however, I found neither. A small mistake, the first paragraph of section 4 appears twice. Also, table 3 is not referred to in the text. But these points do not detract from the work presented, which enriches the area, as could be expected from an author with significant experience in the field.

Everyone with some interest in the area of automatic software repair can find the annotated bibliography useful, but one should keep in mind that the paper does not intend to explain each method in detail. No reference occupies more than eight lines.