No abstract

Training a deep learning model on source code has gained significant traction recently. Since such models reason about vectors of numbers, source code needs to be converted to a code representation and then will be transformed into vectors. Numerous approaches have been proposed to represent source code, from sequences of tokens to abstract syntax trees. However, there is no systematic study to understand the effect of code representation on learning performance. Through a controlled experiment, we examine the impact of various code representations on model accuracy and usefulness in learning-based program repair. We train 21 different models, including 14 different homogeneous code representations, four mixed representations for the buggy and fixed code, and three different embeddings. We also conduct a user study to qualitatively evaluate the usefulness of inferred fixes in different code representations. Our results highlight the importance of code representation and its impact on learning and usefulness. Our findings indicate that (1) while code abstractions help the learning process, they can adversely impact the usefulness of inferred fixes from a developer's point of view; this emphasizes the need to look at the patches generated from the practitioner's perspective, which is often neglected in the literature, (2) mixed representations can outperform homogeneous code representations, (3) bug type can affect the effectiveness of different code representations; although current techniques use a single code representation for all bug types, there is no single best code representation applicable to all bug types. Table 1: Different code representations for the example of Listing 1 Representation ID (RID) Category Representation Example WT1 non-AST based Word tokenization [20, 21, 24] setTimeout ( delay , fn ) WT2 Enhanced word tokenization [27, 41] set Timeout ( delay , fn ) DB1 DeepBugs [58] ID setTimeout ( ID delay , ID fn ) DB2 DeepBugs with types and variable values ID setTimeout ( ID number delay , ID function fn ) DB3 DeepBugs with types without variable values ID setTimeout ( ID number , ID function ) FS1 Function signature setTimeout ( number , function ) FS2 Function signature with position anchors setTimeout ( arg0 number , arg1 function ) FS3 Function signature with LIT/ID setTimeout ( ID number , ID function ) FS4 Function signature with position anchors and LIT/ID setTimeout ( arg0 ID number , arg1 ID function ) SR1 SequenceR [11, 66, 68] setTimeout ( Number_1 , Method_1 )

Abstract-The proliferation of sophisticated web technologies requires efficient tools to evaluate the performance of HTTP traffic under various conditions. In this paper, we present HTTP-Automated Evaluation (HTTP-AE) as a multi-user client-side framework for evaluating HTTP performance. The framework can be used to evaluate mechanisms, which improve HTTP performance. We present several case studies in which HTTP-AE is used to evaluate three HTTP acceleration mechanisms deployed in an emulated satellite system. These case studies show that the framework can be used to test different design aspects that may affect HTTP performance. Hence, by using the proposed framework, one can determine the advantages and limitations of different network design configurations.

Contextual information plays a vital role for software developers when understanding and fixing a bug. Consequently, deep learning-based program repair techniques leverage context for bug fixes. However, existing techniques treat context in an arbitrary manner, by extracting code in close proximity of the buggy statement within the enclosing file, class, or method, without any analysis to find actual relations with the bug. To reduce noise, they use a predefined maximum limit on the number of tokens to be used as context. We present a program slicing-based approach, in which instead of arbitrarily including code as context, we analyze statements that have a control or data dependency on the buggy statement. We propose a novel concept called dual slicing, which leverages the context of both buggy and fixed versions of the code to capture relevant repair ingredients. We present our technique and tool called Katana, the first to apply slicing-based context for a program repair task. The results show Katana effectively preserves sufficient information for a model to choose contextual information while reducing noise. We compare against four recent state-of-the-art context-aware program repair techniques. Our results show Katana fixes between 1.5 to 3.7 times more bugs than existing techniques.CCS Concepts: • Software and its engineering → Software maintenance tools.