Parse notes
Stripe developed a “smart retries” tool that uses ML to determine the best time to retry failed payments. The tool helped some businesses recover over $100 million in revenue. This article goes into detail on how the tool was implemented and provides 4 key lessons on building and choosing the right ML model and giving customers control over how often payments are retried.
The team at SonarQube recently upgraded from React 17 to 18. In this article, they share the three biggest issues and lessons they learned during the process. TypeScript needs to be updated with React. There is a set of codemods released by the React Core team that helps to automatically update types, but they come with a couple of issues. React 18 enables automatic batching by default which can boost performance, but it can also break UI tests and uncover other underlying issues.
Cumulative Layout Shift (CLS) measures how much a page shifts while it is being loaded. It can negatively impact both search rankings and user perception of site speed. Images are a common cause of CLS problems. To improve CLS, web developers can give the browser hints of what size images will be by giving images a width and height attribute separate from the CSS width and height attributes that are part of the style tag.
Scrum is often used as a process to manage teams, especially remote teams. However, it can add 8 hours of meetings per sprint, which is equal to 2 full days of productivity per month. Many projects can’t be easily forced into a 2-week sprint process. Using a collaborative document for each project that addresses the key questions is more efficient. Meetings can be held to address denser tradeoff decisions as long interactions over async communication can create long lead times.
Estimating project timelines in software is hard. Initial estimates are commonly disrupted by external pressures to deliver faster. Estimates are also pressured by cross-functional teams, upper management, and ego-driven desires for faster delivery. To deal with these pressures, it’s helpful to focus on project scope, use productivity metrics, and address optimism bias and black swan events.
Sponsored
Inefficient code, resource bottlenecks, and slow database calls slow your app to a crawl. In this workshop, Sentry [10] engineer Tony Xiao will demo the latest feature he built for Sentry Performance – automatic regression detection for API endpoints [10]. See how to trace end-to-end requests, identify root causes, and nip endpoint
Zed is a high-performance multiplayer code editor from the creators of the code editor Atom.
Retriever lets you request secrets from anyone without any of the data going to a server.
Resend’s React Email 2 includes features such as a new preview experience, significant performance improvements, new components for displaying code, and the ability to deploy to Vercel. The preview experience has been revamped with new mobile previews, a revamped file tree, better user feedback, and improved error handling. New components, including a code-block and code-inline component, make it easier to display code in emails.
Jam, a bug reporting tool, achieved remarkable growth in 2023, going from zero to one million Jams created. It took 18 months and seven failed launches before it achieved product-market fit. To reach this milestone, Jam used internal testing extensively, focused on quality, had a small group of focused pilot users, and tracked retention.
Just like a house needs maintenance, so does software. Software requires ongoing maintenance and support, which varies depending on the model of software (open-source, on-prem, or SaaS). SaaS changed the landscape by offering fixed "rent" costs and reducing the need for in-house support. The “one-off” software model is unsustainable in the long term and it often leads to worse support, buggy software, and a shoddy product.
An entrepreneur gives up after having tried many things, with nothing working.
Fake spinners and loading bars often give users a better user experience by giving them confidence in the result after the progress indicator disappears.
A story about an obscure Apple flag for the DNS resolver on the system that broke one’s Spotify client.
Ollama lets you run LLM models locally. Python and JavaScript libraries are now available for Ollama.