4 ms·
"Attempt 1. Bug dashboard. Initially, in 2006, FindBugs was integrated as a centralized tool that ran nightly over the entire Google codebase, producing a datab
by mcguire 9y ago
"Attempt 1. Bug dashboard. Initially, in 2006, FindBugs was integrated as a centralized tool that ran nightly over the entire Google codebase, producing a database of findings engineers could examine through a dashboard. Although FindBugs found hundreds of bugs in Google's Java codebase, the dashboard saw little use because a bug dashboard was outside the developers' usual workflow, and distinguishing between new and existing static-analysis issues was distracting.
"Attempt 2. Filing bugs. The BugBot team then began to manually triage new issues found by each nightly FindBugs run, filing bug reports for the most important ones. In May 2009, hundreds of Google engineers participated in a companywide "Fixit" week, focusing on addressing FindBugs warnings.3 They reviewed a total of 3,954 such warnings (42% of 9,473 total), but only 16% (640) were actually fixed, despite the fact that 44% of reviewed issues (1,746) resulted in a bug report being filed. Although the Fixit validated that many issues found by FindBugs were actual bugs, a significant fraction were not important enough to fix in practice. Manually triaging issues and filing bug reports is not sustainable at a large scale.
"Attempt 3. Code review integration. The BugBot team then implemented a system in which FindBugs automatically ran when a proposed change was sent for review, posting results as comments on the code-review thread, something the code-review team was already doing for style/formatting issues. Google developers could suppress false positives and apply FindBugs' confidence in the result to filter comments. The tooling further attempted to show only new FindBugs warnings but sometimes miscategorized issues as new. Such integration was discontinued when the code-review tool was replaced in 2011 for two main reasons: the presence of effective false positives caused developers to lose confidence in the tool, and developer customization resulted in an inconsistent view of analysis results."
Once upon a time, I had a co-worker on a C++ project commit a change disabling -Wall with a comment that can loosely be paraphrased as, "No one has time for that."
It's good to see Google has the same issues.