Outline a potential approach

desktop/src/main/log.ts

@@ -32,3 +32,60 @@ export function logErrorSentry(
         console.log(error, { msg, info });
     }
 }
+
+const logError1 = (message: string, e?: unknown) => {
+    if (e === undefined || e === null) {
+        log.error(message);
+        return;
+    }
+
+    let es: string;
+    if (e instanceof Error) {
+        // In practice, we expect ourselves to be called with Error objects, so
+        // this is the happy path so to say.
+        es = `${e.name}: ${e.message}\n${e.stack}`;
+    } else {
+        // For the rest rare cases, use the default string serialization of e.
+        es = String(e);
+    }
+
+    log.error(`${message}: ${es}`);
+};
+
+const logInfo = (message: string) => {
+    log.info(message);
+};
+
+const logDebug = (message: () => string) => {
+    if (isDev) log.debug(() => message);
+};
+
+export default {
+    /**
+     * Log an error message with an optional associated error object.
+     *
+     * {@link e} is generally expected to be an `instanceof Error` but it can be
+     * any arbitrary object too that we obtain, say, when in a try-catch
+     * handler.
+     *
+     * The log is written to disk, and is also printed to the console.
+     */
+    error: logError1,
+    /**
+     * Log a message.
+     *
+     * The log is written to disk, and is also printed to the console.
+     */
+    info: logInfo,
+    /**
+     * Log a debug message.
+     *
+     * The log is not written to disk. And it is printed to the console only on
+     * development builds.
+     *
+     * To avoid running unnecessary code in release builds, this function takes
+     * a function to call to get the log message instead of directly taking the
+     * message. This function will only be called in development builds.
+     */
+    debug: logDebug,
+};