{% /raw %} ## node:vm `cachedData` Support Bun now supports the `cachedData` and `produceCachedData` options in the Node.js `vm.Script` API, along with the `createCachedData` method. This allows for improved performance by caching and reusing compiled JavaScript code. ```js // Create a script with cached bytecode const script = new vm.Script('console.log("Hello world!")', { produceCachedData: true, }); // Later use the cached data const cachedData = script.createCachedData(); // Create a new script with the cached bytecode const newScript = new vm.Script('console.log("Hello world!")', { cachedData: cachedData, }); ``` Thanks to @heimskr for the contribution! ## node:net BlockList support Bun now implements the `BlockList` class from the Node.js `net` module, allowing you to create rules to block specific IP addresses, ranges, or subnets. ```js // Create a new BlockList const blockList = new net.BlockList(); // Add rules blockList.addAddress("123.123.123.123"); blockList.addRange("10.0.0.1", "10.0.0.10"); blockList.addSubnet("8.8.8.8", 24); // Check if an address is blocked blockList.check("123.123.123.123"); // true blockList.check("8.8.8.9"); // true blockList.check("192.168.1.1"); // false ``` Thanks to @nektro for the contribution! ### node:http compatibility improvements This release fixes several edge cases in the `node:http` module, including: - `Agent` abort handling - Extra arguments passed to urls in `http.request` ## More bugfixes ### Fixed: rare crash when importing invalid file URLs Fixed an issue where importing invalid file URLs could sometimes cause Bun to crash. ### Fixed: Erroring in TextDecoder encoding label handling Fixed issues with TextDecoder encoding label handling, bringing implementation in line with the WHATWG spec. Invalid labels containing null bytes now properly throw errors, and TextDecoder.encoding correctly returns the normalized encoding name for all supported encodings. ```js // Previously allowed, now correctly throws: try { new TextDecoder("utf-8\0"); } catch (e) { console.error("Error:", e); // RangeError: ERR_ENCODING_NOT_SUPPORTED } // Now correctly returns the normalized encoding name const decoder = new TextDecoder("utf-16be"); console.log(decoder.encoding); // "utf-16be" (previously returned empty string) ``` Thanks to @pfgithub for the contribution! ### Fixed: TextDecoder 'fatal' option coercion The TextDecoder's `fatal` option now properly coerces values to boolean instead of requiring strict boolean types. This matches browser behavior where values like `0`, `1`, `null`, or strings are automatically converted to appropriate boolean values. ```js // Now works without errors const decoder1 = new TextDecoder("utf-8", { fatal: 1 }); console.log(decoder1.fatal); // true const decoder2 = new TextDecoder("utf-8", { fatal: 0 }); console.log(decoder2.fatal); // false const decoder3 = new TextDecoder("utf-8", { fatal: "any string" }); console.log(decoder3.fatal); // true const decoder4 = new TextDecoder("utf-8", { fatal: null }); console.log(decoder4.fatal); // false ``` Thanks to @albus-droid for the contribution! ### Hot reloading reliability bugfixes A couple bugs have been fixed that could cause `bun --hot` to spuriously crash after a long time. ### Thanks to 10 contributors! - [@190n](https://github.com/190n) - [@albus-droid](https://github.com/albus-droid) - [@alii](https://github.com/alii) - [@dylan-conway](https://github.com/dylan-conway) - [@electroid](https://github.com/electroid) - [@heimskr](https://github.com/heimskr) - [@jarred-sumner](https://github.com/jarred-sumner) - [@nektro](https://github.com/nektro) - [@paperclover](https://github.com/paperclover) - [@pfgithub](https://github.com/pfgithub)In the next version of Bun
— Bun (@bunjavascript) May 4, 2025
98.4% of Node.js' test suite for the "timers" module pass in Bun. pic.twitter.com/YfyyAr8Hc8