11.07.2015 Views

Comparison of Seven Bug Report Types: A Case-Study of ... - IIIT

Comparison of Seven Bug Report Types: A Case-Study of ... - IIIT

Comparison of Seven Bug Report Types: A Case-Study of ... - IIIT

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

TABLE XIIINUMBER OF STARSCRASH REGR. SECUR. CLEAN POLISH PERF. USAB.Min 1 0 0 0 0 0 1Q1 3 1 1 1 1 2 2Median 7 2 1 1 2 2 3Q3 12 5 2 2 3 5 9Max 179 647 59 28 40 78 391TABLE XIVNUMBER OF COMMENTSCRASH REGR. SECUR. CLEAN POLISH PERF. USAB.Min 0 0 1 1 1 1 2Q1 1 1 10 3 5 6 5Median 2 2 15 5 7 8 8Q3 3 5 22 8.5 12 14 26Max 219 647 81 71 76 59 126TABLE XVMILESTONE CHANGECRASH REGR. SECUR. CLEAN POLISH PERF. USAB.Num. BRs at leastone MoveFrom/BulkMove 2137 (56%) 2034 (69%) 37 (9.29%) 151 (18.9%) 165 (63.2%) 93 (55%) 33 (86.8%)Avg. 1.048 1.25 1 1.093 1 1.064 1.121Std. Dev. 0.350 0.209 0 0.420 0 0.383 0.477F. Number <strong>of</strong> Stars and Amount <strong>of</strong> DiscussionIn addition to bug reporter and bug fixer, several developersor users contribute to the threaded discussion and are interestedon activities related to the bug report. We compute number <strong>of</strong>stars and number <strong>of</strong> comments associated to a bug report.Our interest is to measure number <strong>of</strong> developers or personsshowing interest and collaborating for different bug types. InGoogle Chrome project, any user can star 13 an issue indicatingusers interest on a specific issue. The issue tracking systemautomatically sends an email notification (triggered by changein the status <strong>of</strong> the issue) to all the users who starred the issue.We extract both number <strong>of</strong> start and comments for all the bugreports using the programming APIs. Table XIII reveals that ingeneral crash, regression, performance and usability bugsreceive more stars than security, cleanup and polish bugs.The Q2 and Q3 value for number <strong>of</strong> stars for crash bugs ishighest in contrast to the other six types <strong>of</strong> bugs.The Google Issue Tracker provides a facility for users to discussand communicate with each-other by posting comments.The issue tracker is not just a database for archiving bugsbut also serves as an application that acts as a central point<strong>of</strong> communication for the project team [16]. We calculate thenumber <strong>of</strong> comments posted for each bug report and reportthe minimum, Q1, median, Q3 and maximum values in TableXIV. We observe that in general the number <strong>of</strong> commentsper bug report posted for security bug reports is relativelyhigher than other bug reports. The number <strong>of</strong> commentsper bug report (based on Q2 and Q3 values) for crash andregression bug reports is the lowest in contrast to other bugreports. Large number <strong>of</strong> personal interested (large star count13 http://code.google.com/p/support/wiki/IssueTrackeror lots <strong>of</strong> comments) in a bug is an indication <strong>of</strong> bug popularityand may impact bug fixing time [17]. Panjer et al. [17]foundthat bugs having less than four comments (less discussion) getfixed faster as compared to other bugs. Our empirical resultsshows a similar phenomenon; crash and regression have lowerQ2 value for number <strong>of</strong> comments and MTTR as comparedto other bug categories, refer to Table XIV and Table IV.G. Milestone Change FrequencyTable XV shows the absolute numbers and percentage <strong>of</strong>milestone changes for seven different types <strong>of</strong> bug reports.We notice that in general the percentage <strong>of</strong> milestone changesis high (more than 50%). The percentage <strong>of</strong> milestone changesfor crash, regression, polish, performance and usability bug reportsis more than 50%. We did a manual inspection <strong>of</strong> developercomments for bug reports undergoing milestone changeevent to understand the reason behind milestone changes. Wenotice comments mentioning approval <strong>of</strong> milestone changes ifthe reported bug is not a blocker for the currently assignedmilestone. According to the <strong>of</strong>ficial blog 14 by Chromiumproject team, Google Chrome is following the release earlyand release <strong>of</strong>ten policy (shorter release cycle : once everysix weeks). A blog 15 mentions that because <strong>of</strong> the rapidrelease cycle, if a feature is not ready then it is moved to thenext release (new release milestones are created and ticketsare moved). A study by Baysal et al. [18], compares thelifespan <strong>of</strong> major releases <strong>of</strong> FireFox and Chrome browsersand characterizes Chrome as a fast evolving system with short14 http://blog.chromium.org/2010/07/release-early-release-<strong>of</strong>ten.html15 http://blog.assembla.com/assemblablog/tabid/12618/bid/36341/Secrets-<strong>of</strong>-rapid-release-cycles-from-the-Google-Chrome-team.aspx525

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!