New data reveals that fixing bugs and errors in code is the top developer pain point. According to the 2021 Report | The State of Software Code commissioned by Rollbar, manually fixing code is negatively impacting developers’ productivity and morale. At a time when there are not enough developers to meet demand, it appears that more automation is needed to relieve some of the drudgery.
38% of the 950 software developers surveyed said they spend up to a quarter of their time fixing software bugs, with another 26% spending up to half their time. That’s why two-thirds of respondents said they would rather pay bills, visit the in-laws, or go to the dentist before having to face more code fixing.
Yet our long-suffering developers are willing to soldier on due to the vital nature of the work. 89% acknowledged that undetected errors take a big toll on the business. 26% have experienced their companies losing users due to errors in software, and another 26% believe errors damage brand reputation and ability to attract investment.
Despite their best efforts, however, too many bugs and errors slip through the cracks. 62% of respondents had to find out about errors in code from users, 25% heard about them via social media, and 21% had to be told by the CEO. None of those are good ways to hear.
ML to the rescue
Daniel Day, Vice President at Rollbar, touts his company’s Continuous Code Improvement Platform as a way to automate the bug fixing process. Rollbar SDKs capture all exceptions – handled and unhandled – as they occur. They provide diagnostic and contextual data to relieve the pain of resolution.
Competitive platforms such as Sentry and Bugsnag use hard-coded algorithms in their error grouping. Rollbar, on the other hand, harnesses a machine learning grouping engine that groups error patterns and types to reduce noise and enable developers to prioritize those that matter most.
One objection that developers might have with such a tool is that they’ve already built a process to handle errors and issues in their code using a combination of APM, logging, and custom queries. Day says existing processes do not appear to be good enough.
Only the alerts that matter
“When a problem comes up, developers often have so many alerts and logs to dig through to uncover the actual problem,” said Day. “Tedious detective work may be eased a little by filters and custom queries, but it still requires a human to actually look at the field of messages and logs and parse out what is potentially relevant and what is not.”
Rollbar, he said, gives a clear signal that something has gone wrong and helps developers quickly get to the root cause of issues. The goal is to have them deal with errors proactively rather than being caught up in a cycle of endless fixes of the latest error brought to their attention.
With automation entering so many areas of IT, it makes sense that coding would employ more to smooth out its processes. But it remains to be seen whether automated bug fixing platforms can eliminate the grind to such an extent that the process can be consigned to a few minutes at the start or end of the workday. If that can be achieved, then developer productivity will rise sharply across the industry.