- author = "Claire {Le Goues} and Michael Dewey-Vogt and Stephanie Forrest and Westley Weimer",
- title = "A Systematic Study of Automated Program Repair: Fixing 55 out of 105 bugs for \$8 Each",
- booktitle = "34th International Conference on Software Engineering (ICSE 2012)",
- year = "2012",
- editor = "Martin Glinz",
- pages = "3--13",
- address = "Zurich",
- month = jun # " 2-9",
- keywords = "genetic algorithms, genetic programming, genetic improvement, APR, GenProg, algorithmic improvement, automated program repair, cloud computing resource, defect repair, grounded human-competitive cost measurement, off-the-shelf C program, open-source program, program bug, real-world program, repair cost, systematic evaluation, C language, cloud computing, program debugging, public domain software, software cost estimation, software maintenance",
- ISSN = "0270-5257",
-
URL = "
http://dijkstra.cs.virginia.edu/genprog/papers/weimer-icse2012-genprog-preprint.pdf",
-
code_url = "
https://squareslab.github.io/genprog-code/",
-
DOI = "
doi:10.1109/ICSE.2012.6227211",
- size = "11 pages",
- abstract = "There are more bugs in real-world programs than human programmers can realistically address. This paper evaluates two research questions: What fraction of bugs can be repaired automatically? and How much does it cost to repair a bug automatically? In previous work, we presented GenProg, which uses genetic programming to repair defects in off-the-shelf C programs. To answer these questions, we: (1) propose novel algorithmic improvements to GenProg that allow it to scale to large programs and find repairs 68percent more often, (2) exploit GenProg's inherent parallelism using cloud computing resources to provide grounded, human competitive cost measurements, and (3) generate a large, indicative benchmark set to use for systematic evaluations. We evaluate GenProg on 105 defects from 8 open-source programs totalling 5.1 million lines of code and involving 10,193 test cases. GenProg automatically repairs 55 of those 105 defects. To our knowledge, this evaluation is the largest available of its kind, and is often two orders of magnitude larger than previous work in terms of code or test suite size or defect count. Public cloud computing prices allow our 105 runs to be reproduced for 403 USA dollars; a successful repair completes in 96 minutes and costs $7.32, on average.",
-
notes = "GenProg >> ClearView, AutoFix-E, AFix.
Bug bounties, Tarsnap.
Chrome is ordered list of AST edit operations. Delete, insert, swap, uniform crossover. 10 GP runs (population 40, <= ten generations, <12 hours) in parallel on Amazon EC2 (c1.medium, 1.7Gbyte RAM) cloud. Less than 10percent of children fail to compile.
fbc, gmp, gzip, libtiff, lighttpd, php, Python, wireshark approx 5.1 million lines of C code.
Human difficulty of bugfix != GP difficulty?
See also \cite{Fry:2012:ISSTA}
http://www.ifi.uzh.ch/icse2012/
Bronze winner 2012 HUMIES GECCO 2012.
cited by \cite{Nguyen:2013:ICSE} \cite{Qi:2015:APP:2771783.2771791} \cite{DBLP:journals/corr/DurieuxMMSX15}
Also known as \cite{6227211}",
