Skip to main content

Framework

Unreleased#

v5.0.0#

API: Reading, writing with standard streams#

  • Implement app.readProcessInput(readAll=false) for reading string data from the standard input stream. This function can read a single line or multiple lines at once.
  • Implement app.writeProcessOutput and app.writeProcessError for sending string data to standard output and error streams accordingly.

API: filesystem#

  • Enable default recursive behavior in the filesystem.createDirectory function.
  • Add filesystem.copy, filesystem.move, and filesystem.remove functions.
  • Add {recursive} options object to filesystem.readDirectory(path, options) activate recursive directory listing.

API: os#

  • Accept stdin with the background mode of the os.execCommand function.

Core: security#

  • Now, all app clients and extensions require a valid connect token (A part of the NL_TOKEN) to connect to the Neutralinojs framework WebSocket server to receive app events.
  • The framework now sends the access token, connect token, port, and extension identifier to extension processes via the standard input, so other processes can't read them by scanning the process list.
  • The WebSocket client connection origin is checked during the HTTP/WS handshake to make sure that only local clients connect with a specific Neutralinojs app that runs on window, browser, or Chrome mode. Any URL origin is allowed for the cloud mode.
  • The above security enhancements fix two security advisories on GitHub.

Global variables#

  • Now the NL_ARCH global variable returns the same value that computer.getArch returns.

Deprecations#

  • Removed filesystem.removeFile, filesystem.removeDirectory, filesystem.copyFile, and filesystem.moveFile functions.
  • Extensions won't receive --nl-port, --nl-token, and --nl-extension-id as command-line arguments. Now the framework sends these via stdin as a JSON string.

v4.15.0#

Configuration: custom user agent string#

Developers sometimes use the user agent string to indentify the client in server-side and client-side source codes. Now, Neutralinojs lets app developers extend the default user agent string with a custom string via the window.extendUserAgentWith configuration property and the --window-extend-user-agent-with=<string> command-line option, as shown in the following example:

"window": {
// ---
"extendUserAgentWith": "MyAppClient"
}

The above configuration extends the user agent string:

console.log(navigator.userAgent) // <Default UA> MyAppClient

Configuration: custom configuration files#

The Neutralinojs framework typically loads the application configuration content from the neutralino.config.json file, but sometimes app developers need to use a custom configuration filename during development (i.e., For separating production and development environments). Now, you can use the --config-file=<filename> internal command-line parameter to use a custom app configuration, as shown in the following example:

./myapp-linux_x64 --load-dir-res --config-file=neutralino-dev.config.json

v4.14.1#

DevOps#

  • Use Windows 2019 server and macOS 11 (Big Sur) to compile framework release artifacts.
  • Set minimum supported macOS version via MACOSX_DEPLOYMENT_TARGET=10.7

v4.14.0#

Filesystem API#

  • Introduce the filesystem.getWatchers function to get all created watchers.
  • Add binary file reading support for the filesystem.updateOpenedFile(id, action, data) function. This function implements readBinary and readAllBinary actions and dataBinary output action in the openedFile global event.
  • Return the existing watcher identifier from filesystem.createWatcher if there is an existing active watcher for the given path.

OS API#

  • Add a way to set current working directory for process creation functions: os.execCommand(command, options) now supports cwd via the options object and os.spawnProcess(command, cwd) accepts currently working directory via the second string parameter.

Improvements/bugfixes#

  • Fix Unicode charactor issues in the tray menu on Windows.
  • Avoid including null bytes to file reader events initiated by the filesystem.openFile function.
  • Discard window method executions on non-window modes.

DevOps#

  • Update external GitHub Action packages to their latest versions.
  • Fix issues of the failing test cases

v4.13.0#

Core: persistent window state#

Now the framework stores the primary window state in a temporary file (JSON formatted) and loads during the startup process. This feature stores and sets the window position (x, y coordinates), size (width and height), and maximized status. This feature is enabled by default in all platforms, but app developers can turn it off by using false for the window.useSavedState configuration attribute or --window-use-saved-state internal CLI argument.

Also, the window state loading status is available via the NL_WSAVSTLOADED global variable. The framework sets true for this boolean variable if the window state was loaded from the saved configuration.

Improvements/bugfixes#

  • Fix several issues in the webview Windows code (i.e., Wait for the window close event, window style fixes, etc.).

v4.12.0#

API: window#

  • Add window.center to center the application window programmatically.

Core: configuration#

  • Setting the initial window position via modes.window.x and modes.window.y integer config props.
  • Centering the window at startup via modes.window.center boolean config props.
  • Add config overrides for the startup window position: --window-center=<true|false>, --window-x=<int> and --window-y=<int>

Core: webview#

  • Statically link the Webview2 loader library on Windows, so app developers can package their apps without including the WebView2Loader.dll file.

Improvements/bugfixes#

  • Added Unicode characters support for Neutralinojs Windows.

v4.11.0#

API: File watchers#

In some scenarios, Neutralinojs app developers need to implement file watchers in their apps. Earlier, Neutralinojs API didn't offer a native file watcher API and developers had to use less-performant workarounds, such as making recursive filesystem.readDirectory calls. The new file watchers API lets you create native, cross-platform, event-based file watchers for filesystem paths with the following functions:

  • filesystem.createWatcher(path): Creates a new file watcher for a given path and returns the watcher identifier.
  • filesystem.removeWatcher(watcherId): Removes a files watcher based on a file watcher identifier.

Whenever a filesystem change occurs, file watcher instances dispatch the watchFile event with the following data:

  • id: Watcher identifier
  • action: Filesystem change: add, delete, modified, and moved
  • dir: Directory path
  • filename: Modified file

Improvements/bugfixes#

  • Fix the initial window flashing and hidden window state issues on Windows.
  • Fix data corruption in binary file appending.
  • Apply dark/light themes based on system preferences on Windows.

v4.10.0#

DevOps#

  • Generate official macOS arm64 (For Apple M1 and higher devices) and universal binaries with the GitHub workflow.

Improvements/bugfixes#

  • Add a JSON schema definition for the neutralino.config.json file.
  • Fix source code compilation issues on macOS arm64 systems.

ESM/NPM support#

Neutralinojs typically stores the client library implementation in a separate file (neutralino.js). This approach makes Neutralinojs app development process easier with a globally exposed JavaScript object, Neutralino. But, modern web developers use ES modules and they usually like to fetch dependencies from NPM. So, now, developers can load the __neutralino_globals.js internal script (i.e., <script src="__neutralino_globals.js"></script>) to load only globals. Then, they can use the client library implementation via neutralino.mjs with the neu CLI or from @neutralinojs/lib with a Node package manager.

v4.9.0#

API: Custom methods#

Neutralinojs offers the extensions API to write custom backend code with any programming language, but extensions come with the following drawbacks that affect apps in several scenarios:

  • Extensions use a shared WebSocket for communication, so using direct C++ references (i.e., the window handler) is impossible within extensions.
  • The developer is responsible for packaging their extension binaries.
  • A C++-based extension is not fast as native C++-based code due to the WebSockets-based IPC.

Alternatively, a developer can download the framework C++ code, modify it, and re-compile it. But, the developer may face issues while synching upstream code modifications. So, Neutralinojs offers a separate namespace, a function template, inbuilt helper functions (i.e., to get the window handler, validation, etc.), and a developer guide to add custom APIs to the Neutralinojs framework without updating the framework core.

Example:

let res = await Neutralino.custom.fetch('https://neutralino.js.org');

If developers make a custom API that others can use, we motivate them to contribute to the Neutralinojs framework by adding it to the main codebase.

Example:

let res = await Neutralino.net.fetch('https://neutralino.js.org');

If the developer adds a new custom method to the framework, the client library will automatically export it to the Neutralino.custom namespace by using the NL_CMETHODS internal global variable. The Neutralino.custom.getMethods function returns an array of custom methods similar to NL_CMETHODS.

API: File streams#

The current Neutralinojs API offers non-stream-based (Promise-based but synchronous-like) functions for working with files. For example, the filesystem.readFile function reads the file content and retrieves data synchronously even though the WebSocket communication mechanism supports asynchronous patterns. However, the synchronous API lets developers work with files in a simple way, but they face the following issues in some scenarios:

  • Reading large files is not performance-friendly (The whole file content gets loaded into the memory).
  • Unable to work with dynamic file objects (i.e., Device files).
  • File descriptors are not persistent within the app lifecycle as they get destroyed after native API calls.

To solve this issue, we offer an event-based file stream API with the following functions/events:

Functions#

  • filesystem.openFile: Creates a file stream by openning a file.
  • filesystem.updateOpenedFile: Triggers a file read/readAll event or sets the file cursor.
  • filesystem.getOpenedFileInfo: Returns (awaited) information about the file stream (Props: id, eof, pos, and lastRead)

Events#

  • openedFile: Occurs per each file read event and whenever the file stream reaches EOF.

v4.8.0#

API: os.getEnvs#

os.getEnv returns a value for a given environment variable key. Developers had to use alternative methods to retrieve a list of all environment variables with values. The os.getEnvs returns all environment variables as a JavaScript object similar to Node's process.env.

API: filesystem#

  • Added pos and size options for readFile and readBinaryFile methods to set the file cursor position and buffer size respectively.

API: storage.getKeys#

Added the storage.getKeys function to get an array of Neutralinojs storage keys. Now, developers don't need to write their own functions to retrieve storage keys with the filesystem API.

API: computer.getMousePosition#

Returns the current mouse cursor position via a JavaScript object that has x and y props. This function is helpful for develping interactive desktop widgets on all supported platforms.

Bugfixes/improvements#

  • Replaced string error codes with enums in the C++ source code.
  • Some refactorings done in C++ struct definitions and return values.

v4.7.0#

API: System information API#

Ealier, we had the getMemoryInfo function in the computer namespace to retrieve system memory statistics. Now, we have added more functions to get details about the CPU, operating system, kernel, and connected displays:

  • computer.getArch: Returns the CPU architecture. i.e, x64, arm, etc.
  • computer.getKernelInfo: Returns the operating system's kernel details.
  • computer.getOSInfo: Returns the operating system details.
  • computer.getCPUInfo: Returns the CPU details.
  • computer.getDisplays: Returns an array of all connected displays with the resolution, frequency-like information.

API: os dialogs#

  • Add the defaultPath option to showSaveDialog, showOpenDialog, and showFolderDialog functions to set the initial path/filename of system dialogs.

DevOps#

  • Run the test suite on Windows GitHub Actions instance.
  • Fix the armhf framework binary generation issue in the GitHub Actions workflow.

Bugfixes/improvements#

  • Hide the automation info bar from the Chrome mode by default -- Developers can add the --enable-automation flag to the config file if they need the particular command-line switch.

v4.6.0#

API: Process spawning API#

We have os.execCommand for launching processes, but it's synchronous, meaning, the developer has to wait unti process completion to receive pid, stdOut and stdErr. execCommand is not suitable for long-running processes. The new spawning API offers API functions for handling long-running processes in a multi-threaded way.

  • os.spawnProcess(command): Spawns a process and returns id (A virtual Neutralino-scoped pid) and pid (Operating system-level pid).
  • os.getSpawnedProcesses(): Returns a list of spawned processes.
  • os.updateSpawnedProcess(id, action, data): Sends an action event for the spawned process. Supports the following actions:
    • stdIn: Sends a string via the standard input stream. data is the input string.
    • stdInEnd: Closes the standard input stream.
    • exit: Terminates the spawned process.

Core: events#

  • spawnedProcess: Dispatched when there is a change in the spawned process. CustomEvent gets triggered with the following object:
{
id: <id>,
pid: <pid>,
action: <action>,
data: <data>
}

Available actions:

  • stdOut: Outputs standard output data. data contains the standard output payload.
  • stdErr: Outputs standard error data. data contains the standard error payload.
  • exit: Notified when the process terminates. data contains the process exit code.

DevOps#

  • Add official Linux ARM (armhf and arm64) binary build support for the standard release workflow and nightly build.

API: filesystem#

  • Expose createdAt and modifiedAt JavaScript timestamps via the filesystem.getStats function.

Bugfixes/improvements#

  • Hide the .storage directory (Neutralinojs storage location) on Windows.
  • Support using query parameters in resources URLs. i.e., index.html?v=2.0 is supported, but caching is not yet implemented.
  • Show a user-friendly error message for Neutralinojs server initialization failures.

v4.5.0#

Core: events#

  • Added windowFocus and windowBlur native events.

DevOps#

  • Use BuildZri for C++ build automation.
  • Implement a new workflow for nightly releases. Developers can get the nightly release binaries via the nightly version tag.

Bugfixes/improvements#

  • Search and dynamically load app indicator libraries on Linux. This enhancement supports Neutralinojs to start without a crash when there is no app indicator library present on the system.
  • Throw NE_OS_TRAYIER from the os.setTray function for initialization failures.
  • Fix binary file write error on Windows with the filesystem.writeBinaryFile function. This was fixed by using LF as the line breaker on all platforms as a portable solution. If the developer needs CRLF on Windows (or CR on Darwin), the developer needs to handle it explicitly with NL_OS.

Core: global variables#

  • Add NL_COMMIT to hold framework's release commit. This is helpful to find the nightly release's code snapshot.

v4.4.0#

API: window#

  • Add window.getPosition to get the current window coordinates.

API: filesystem#

  • Add filesystem.appendFile to append text content to a file. Thrown errors are similar to the filesystem.writeFile function.
  • Add filesystem.appendBinaryFile to append binary content to a file. Thrown errors are similar to the filesystem.writeBinaryFile function.

Dependencies#

  • Use ayatana-appindicator3-0.1 instead of appindicator3-0.1 (Marked as obsolete in Debian packages)

Bugfixes/improvements#

  • Dynamically call SetProcessDpiAwarenessContext function to support previous Windows versions.
  • Display the native method name with the NE_RT_NATPRME error payload.

v4.3.0#

Core: auth#

  • Added tokenSecurity to the configuration to improve the security of NL_TOKEN. Accepted values are none and one-time. If none is set, Neutralinojs server will always expose NL_TOKEN for any application instance, so you can open Neutralinojs apps from browser directly. If one-time (recommended) is set, Neutralinojs server expose NL_TOKEN only once and client persists the token in sessionStorage, so all other clients that acces the app after the initial client will get an auth error message (Displayed to the user via HTML).

API: window#

  • window.setAlwaysOnTop(bool) was added.
  • window.getSize was added.

Core: global variables#

  • Added NL_EXTENABLED which returns true if extensions are enabled. It is used to check extensions status faster during initialization process.

Bugfixes#

  • Fixed the file dialog order issue on Windows.
  • Removed tray icon when the application exits.

v4.2.0#

API: clipboard#

  • Add clipboard API for all platforms. Supports reading and writing text in the system clipboard.

Core: configuration#

  • Accept chrome-related CLI args: --chrome-width, --chrome-height, and --chrome-args.
  • Accept --neu-dev-extension to enable devtools connections (For internal usages with CLI).
  • Accept custom HTTP headers via serverHeaders option.

API: global variables#

  • Added NL_RESMODE to get details about application resources source. Returns bundle if resources are loaded from resources.neu. Otherwise, returns directory.

Bugfixes#

  • Fix chrome mode user data dir issue (Support paths with spaces).
  • Fix browser mode's shutdown issue with neu CLI.

v4.1.0#

API: window#

  • window.getTitle returns the current native window title.

API: os#

  • Remove NE_OS_ENVNOEX from os.getEnv and return an empty string if environment variable is not defined.

Core: Chrome mode#

  • Now you can run Neutralinojs apps as Chrome apps via the chrome mode.

Core: configuration#

  • Getting chrome mode config from modes.chrome. Added height, width, and args specially for the chrome mode.

v4.0.0#

API: os.execCommand#

  • Renamed shouldRunInBackground option to background.
  • Supports stdIn as an optional input.
  • Returns multiple values: stdOut, stdErr, pid, and exitCode.

Core: API extensions#

  • Config file supports new extensions array globally or in a specific mode.
  • Neutralino spawns extension processes based on extensions array.
  • Neutralino doesn't send kill signals to extension processes. The extension developer needs to stop processes properly.
  • extension setting has 4 fields: id, command, commandLinux,commandDarwin, and commandWindows.
  • The command's (platform-specific command's) value accepts ${NL_PATH} global variable.
  • Each extension process instance is called with three CLI args: --nl-port={}, --nl-extension-id={} and --nl-token={}
  • Enable/disable extensions with enableExtensions config.

Core: Static server#

  • Supports setting document root (with documentRoot) via config. Now, it's possible to launch app without a subdirectory in the URL.

Core: Community driver processes#

Now developers can use Neutralinojs as a part of their software with any programming language by spawning Neutralinojs as a child process. Use exportAuthInfo to write auth details to ${NL_PATH}/.tmp/auth_info.json. Then the parent process can pickup access details there. Note that WebSocket communication needs to be initiated via extensions API/loader.

API: app.broadcast#

  • Sends an event to all app clients. This method is for extension developers.

API: events.broadcast#

  • Sends an event to all clients (apps and extensions). Useful for notifying important general events.

API: extensions#

  • dispatch: Sends an event to a specific extension.
  • broadcast: Sends an event to all connected extensions. Useful for sending shutdown signals.
  • getStats: Returns details about loaded extensions and connected extensions.

API: updater#

  • checkForUpdates: Send a request to a seed URL (JSON) and fetch update details.
  • install: Install updates based on the currently downloaded manifest.

Events#

  • appClientConnect and appClientDisconnect: Occurs when a new app instance is launched and closed respectively.
  • extClientConnect and extClientDisconnect: Occurs when a new extension is connected and disconnected respectively.
  • extensionReady can be used to implement immediate extension calls. This is implemented from the client-side with extensions.getStats and extClientConnect. This event gurantees that it will be triggered regardless of the extension's start time.

Error codes#

  • NE_EX_EXTNOTC: Thrown by extensions.dispatch if the target extension is not connected.
  • NE_UP_CUPDMER: Thrown by updater.checkForUpdates if the JSON update manifest is invalid or applicationId is not matching.
  • NE_UP_CUPDERR: Thrown by updater.checkForUpdates if the updater API cannot fetch the manifest.
  • NE_UP_UPDNOUF: Thrown by updater.install when the update manifest is not loaded.
  • NE_UP_UPDINER: Thrown by updater.install for update installation errors.

Bug fixes#

  • Fix port 0 issue with modes. Earlier, if the developer sets port as 0 from a specific mode, the NL_PORT also becomes 0.
  • Fix an issue with writeToLogFile config option. Earlier, the log file was created even this option is set to false.

Global variables#

  • NL_APPVERSION: Value of the version key in the config file.