January 2016 Summaries

The Sentry Java SDK is a widely adopted logging framework that provides integration with the popular JVM dependency managers Maven and others, making it easy to choose the right one for your application. The SDK offers several logging integrations, including logback, Log4j, and Log4j 2, which can be used to send events to a Sentry server, providing a unified logging API that can be used with any backing implementation. When choosing a logging framework, developers should consider the ease of use, control over reporting configuration, and compatibility with other libraries in their application. The Sentry Java SDK also provides additional features such as HTTP request details, mapped diagnostic context, asynchronous logging, and support for multiple JVM languages including Clojure, Scala, Kotlin, and Groovy.

The new Saved Searches feature has been introduced by Sentry, replacing predefined filters with a dropdown menu. The default searches include New Today, Needs Triage, Unresolved Issues, Assigned to Me, and My Bookmarks. Admins can customize the team default and their personal default, which can be easily created and saved using the Saved Search selector. The feature will continue to evolve based on user feedback, with plans to expose a more accessible UI for non-admins and potentially allowing users to create unique saved searches that aren't shared by default.

Sentry aims to make crash reporting pleasant by providing a general bill of health for itself through metrics capture and system-level monitoring. However, logs remain a weak point due to sensitive data concerns, prompting the company's focus on unified logging structures and enabling container log shipping. Uptime is another major concern, with Sentry relying on external services like New Relic and Pingdom for validation. The company seeks to improve its tooling in various areas while maintaining flexibility for existing infrastructure and customer installations behind firewalls.

Theresa Vermeersch is joining our team, bringing a unique blend of interests and hobbies outside of work. We're excited to have her on board and encourage everyone to say hello on Twitter! As an avid coffee connoisseur, book nerd, and art enthusiast, Theresa can be found enjoying the outdoors or perfecting recipes in her free time.

Sentry 8 On-Premise has been formally shipped to the Python Package Index after three months of improvement and refinement. The major features introduced in Sentry 8 include a new configuration format, config.yml, as well as a simplified configuration flow for new installs and upgrades. The organization dashboard has also been revamped with features such as assigned issues views, bookmarks, and history, providing users with a quick overview of their projects. Additionally, the backend has undergone significant overhauls, including the expansion of the API and the introduction of a simple way to import and export critical system data. Users are encouraged to review the changelog before upgrading to ensure compatibility, as MySQL is no longer supported for on-premise installations.

The Sentry project has shifted from a server-side rendering approach to a rich React-powered user interface rendered purely on the client side. To solve the client-side i18n problem, they integrated their existing translation pipeline with a custom solution that matches the client and server behavior of gettext. The team had to overcome complexities such as formatting placeholders, handling complex interactions like links, and extracting translations from JavaScript code. They developed a custom Babel plugin to automate this process and created a custom webpack loader to ship translations with the frontend app. To improve their context support and ensure all strings get correctly marked for translation, they plan to enhance their tooling in the future.

The window.onerror event is a special browser event that fires whenever an uncaught JavaScript error has been thrown. It's one of the easiest ways to log client-side errors and report them to your servers. By assigning a function to window.onerror, you can listen to this event and handle errors accordingly. The Error object and its stack property are particularly valuable for debugging purposes as they provide detailed information about the source location of each frame of the program when the error occurred. However, the implementation of the stack property differs among browsers, so tools like TraceKit, stacktrace.js, etc., can be used to normalize error strings across different browsers. The window.onerror event has been available in browsers for some time and is supported by most modern browsers. To capture as much error information as possible, you can also wrap functions in try/catch statements. Finally, the captured error data needs to be transmitted to your servers using a reporting web service that supports CORS if necessary.