How should compilers explain problems to developers?

This week I presented the final part of my dissertation, How should compilers explain problems to developers?, at the Foundations of Software Engineering conference in Lake Buena Vista, Florida. This work is relevant to communities, such as Elm and Rust, that are invested in improving the usability of the error messages their tools produce.

The abstract of the paper follows:

Compilers primarily give feedback about problems to developers through the use of error messages. Unfortunately, developers routinely find these messages to be confusing and unhelpful. In this paper, we postulate that because error messages present poor explanations, theories of explanation—such as Toulmin’s model of argument—can be applied to improve their quality. To understand how compilers should present explanations to developers, we conducted a comparative evaluation with 68 professional software developers and an empirical study of compiler error messages found in Stack Overflow questions across seven different programming languages.

Our findings suggest that, given a pair of error messages, developers significantly prefer the error message that employs proper argument structure over a deficient argument structure when neither offers a resolution—but will accept a deficient argument structure if it provides a resolution to the problem. Human-authored explanations on Stack Overflow converge to one of the three argument structures: those that provide a resolution to the error, simple arguments, and extended arguments that provide additional evidence for the problem. Finally, we contribute three practical design principles to inform the design and evaluation of compiler error messages.

One λ at a time

Lambda
Our full paper, One λ at a time: What do we know about presenting human-friendly output from program analysis tools?, has been accepted to the 8th Workshop on Evaluation and Usability of Programming Languages and Tools (PLATEAU) at SPLASH 2017, Vancouver, British Columbia. The paper is useful for human-computer interaction researchers who want to understand the design space of error messages from program analysis tools.

The abstract of the paper follows:

Program analysis tools perform sophisticated analysis on source code to help programmers resolve compiler errors, apply optimizations, and identify security vulnerabilities. Despite the utility of these tools, research suggests that programmers do not frequently adopt them in practice—a primary reason being that the output of these tools is difficult to understand. Towards providing a synthesis of what researchers know about the presentation of program analysis output to programmers, we conducted a scoping review of the PLDI conference proceedings from 1988-2017. The scoping review serves as interim guidance for advancing collaborations between research disciplines. We discuss how cross-disciplinary communities, such as PLATEAU, are critical to improving the usability of program analysis tools.

Expressions on the Nature and Significance of Programming and Play

All is play.

My single-author full paper, Expressions on the Nature and Significance of Programming and Play, has been accepted to the IEEE Symposium on Visual Languages and Human-Centric Computing. I presented the paper in Raleigh, North Carolina.

The abstract of the paper follows:

Play is all around us, an essential and innate phenomenon that serves as an important mediator in creativity, interest, learning, and drive. Though play is thought to be universal, the way in which it materializes is situationally-dependent and not well-understood, particularly in software engineering. To understand how programmers express the concept of play, we conducted a qualitative study on the online social news website, Hacker News—a venue for software practitioners. From Hacker News, we qualitatively analyzed nearly 1,000 user-submitted comments containing the terms “programming” and “play.” The contribution of this work is a contemporary synthesis of how software practitioners interpret programming and play in experiential terms. Our findings suggest how programming and play can be understood through rich metaphors, among them, play as: art, playgrounds, spontaneity, and tinkering. Hacker News authors reflect about childhood experiences as a catalyst for learning programming, and contrast play against work.

Microsoft in Redmond

I’m delighted to announce that as of today, I’ve started the next phase of my research career at Microsoft. I’ll be working with the PROSE team in a dual role: as a Researcher at the intersection of programming languages and human-computer interaction, and as a Research Software Engineer to improve developer experiences across Microsoft’s software development tools.

Do Developers Read Compiler Error Messages?


Our full paper, Do Developers Read Compiler Error Messages?, has been accepted to the International Conference on Software Engineering. I will be presenting the paper at the conference in Buenos Aires, Argentina.

The abstract of the paper follows:

In integrated development environments, developers receive compiler error messages through a variety of textual and visual mechanisms, such as popups and wavy red underlines. Although error messages are the primary means of communicating defects to developers, researchers have a limited understanding on how developers actually use these messages to resolve defects. To understand how developers use error messages, we conducted an eye tracking study with 56 participants from undergraduate and graduate software engineering courses at our university. The participants attempted to resolve common, yet problematic defects in a Java code base within the Eclipse development environment. We found that: 1) participants read error messages and the difficulty of reading these messages is comparable to the difficulty of reading source code, 2) difficulty reading error messages significantly predicts participants’ task performance, and 3) participants allocate a substantial portion of their total task to reading error messages (13%-25%). The results of our study offer empirical justification for the need to improve compiler error messages for developers.

FSE Student Research Competition: How should static analysis tools explain anomalies to developers?

I presented my thesis proposal work, “How should static analysis tools explain anomalies to developers?”, at the Foundations of Software Engineering Student Research Competition in Seattle, Washington. I was awarded second place in the competition. The abstract of the short paper follows:

Despite the advanced static analysis tools available within modern integrated development environments (IDEs) for detecting anomalies, the error messages these tools produce to describe these anomalies remain perplexing for developers to comprehend. This research postulates that tools can computationally expose their internal reasoning processes to generate assistive error explanations that more closely align with how developers explain errors to themselves. My work demonstrates that tools stand to significantly benefit if they incorporate explanation principles in their design.

The associated poster for the paper is also available.

Software Engineering Research for the Post-apocalypse

Designing for Dystopia

I presented our short paper, Designing for Dystopia: Software Engineering Research for the Post-apocalypse, at the Visions and Reflections Track in Foundations of Software Engineering conference at Seattle, Washington.

The abstract of the paper follows:

Software engineering researchers have a tendency to be optimistic about the future. Though useful, optimism bias bolsters unrealistic expectations towards desirable outcomes. We argue that explicitly framing software engineering research through pessimistic futures, or dystopias, will mitigate optimism bias and engender more diverse and thought-provoking research directions. We demonstrate through three pop culture dystopias, Battlestar Galactica, Fallout 3, and Children of Men, how reflecting on dystopian scenarios provides research opportunities as well as implications, such as making research accessible to non-experts, that are relevant to our present.

Check it out before the world ends.

From Quick Fixes to Slow Fixes

Our full paper, From Quick Fixes to Slow Fixes: Reimagining Static Analysis Resolutions to Enable Design Space Exploration, has been accepted to the International Conference on Software Maintenance and Evolution (ICSME). In addition, I also served as Web Chair of the Organizing Committee. ICSME is hosted this year in Raleigh, North Carolina.

The abstract of the paper follows:

Quick Fixes as implemented by IDEs today prioritize the speed of applying the fix as a primary criteria for success. In this paper, we argue that when tools over-optimize this criteria, such tools neglect other dimensions that are important to successfully applying a fix, such as being able to explore the design space of multiple fixes. This is especially true in cases where a fix only partially implements the intention of the developer.

In this paper, we implement an extension to the FindBugs defect finding tool, called FixBugs, an interactive resolution approach within the Eclipse development environment that prioritizes other design criteria to the successful application of suggested fixes. Our empirical evaluation method of 12 developers suggests that FixBugs enables developers to explore alternative designs and balances the benefits of manual fixing with automated fixing, without having to compromise in either effectiveness or efficiency. Our analytic evaluation method with six usability experts identified trade-offs between FixBugs and Quick Fix, and suggests ways in which FixBugs and Quick Fix can offer complementary capabilities to better support developers.

Supplemental materials for the FixBugs study have also been made available.

Blending Programming Environments and Games

Titus presents. Photo by Austin Henley.

I presented our full paper, A Perspective on Blending Programming Environments and Games: Beyond Points, Badges, and Leaderboards, at the IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC 2016) in Cambridge, UK.

The abstract of the paper follows:

Programming environments and game environments share many of the same characteristics, such as requiring their users to understand strategies and solve difficult challenges. Yet, only game designers have been able to capitalize on methods that are consistently able to keep their users engaged. Consequently, software engineers have been increasingly interested in understanding how these game experiences can be transferred to programming experiences, a process termed gamification.

In this perspective paper, we offer a formal argument that gamification as applied today is predominately narrow, placing emphasis on the reward aspects of game mechanics at the expense of other important game elements, such as framing. We argue that more authentic game experiences are possible when programming environments are re-conceptualized and assessed as holistic, serious games. This broad gamification enables us to more effectively apply and leverage the breadth of game elements to the construction and understanding of programming environments.

European Conference on Object-Oriented Programming Retrospective

Through my research in HCI and software engineering tools, I typically make the annual rounds in conferences such as ICSE, FSE, and VL/HCC. Thus, having the opportunity to attend ECOOP 2016 was definitely outside of my comfort area, but it was worthwhile in that it provided me with exposure to an otherwise enigmatic area of Computer Science.

As a first-timer to ECOOP, I expected all of the conference to basically be about programming language (PL) theory (or as we call it in HCI, the land of Greek letters). I was surprised, however, to find workshops on the usability aspects of programming languages. I attended two such workshops:

  • LIVE, a workshop on live programming systems that “abandon the traditional edit-compile-run cycle in favor of fluid user experiences.”
  • Grace, a workshop on the emerging Grace programming language. Originating at ECOOP in 2010, the Grace programming language is designed to allow novices to discover object-oriented programming in simpler ways.

To gain exposure to new ideas, I also attended ICOOOLPS, a workshop on compiler optimization and performance for object-oriented programming.

Although my own research community and ECOOP have relatively little intersection, through these smaller workshops I quickly met new colleagues, including James Noble, Michael Kölling, and my own PhD advisor’s advisor, Andrew Black.

An aspect of ECOOP that I particularly appreciated was the morning breakfast sessions, where students like myself were paired with faculty members to learn more about ECOOP research. I took full advantage of these sessions and introduced myself to a new faculty member each day for breakfast: Matthias Felleisen, Tobias Wrigstad, Laurence Tratt, and Jan Vitek.

Another highlight of the conference was the ECOOP Summer School. The lecturers for these talks made a significant effort to provide a gentle introduction to programming language theory and to explain the types of problems researchers in PL study. One of the more memorable lectures was a hands-on session by Laurence Tratt and Carl Friedrich Bolz, where we worked on our laptops to implement a JIT in Python.

ECOOP Summer School

Thanks again to the NSF for providing this amazing opportunity.