Explore TensorFlow 1.15, Firefox WebSocket Inspector, and Microsoft DAPR & OAM

TensorFlow 1.15 has been officially released as the final minor version of the 1.x branch. No new features will be added to 1.x, and future major functionality will appear in TensorFlow 2.0, scheduled for release at the end of September. Google will continue to provide security patches for at least one year.

Key features in TensorFlow 1.15

The release includes the fully‑implemented compat.v2 and compat.v1 modules, allowing code to emulate TensorFlow 2.0 behavior via enable_v2_behavior(). Standard GPU support is now bundled in the pip package, and the tensorflow-gpu package remains for supported Windows, Linux, and non‑GPU devices.

Notable changes include the split into two pip packages: tensorflow_core (the code) and a virtual package tensorflow. The build uses devtoolset‑7 (GCC 7) on Ubuntu 16, which may affect extension compatibility.

Firefox 70 adds a WebSocket inspector

Firefox Developer Tools will soon feature a new WebSocket inspector, included in the upcoming version 70. The tool helps developers monitor persistent WebSocket connections, which provide a continuous, low‑overhead channel between client and server compared to traditional HTTP polling.

While the inspector is part of the Network panel and can filter open connections, it currently cannot display real‑time WebSocket frames. Future releases aim to add binary payload viewers, closed‑connection displays, and export capabilities.

Windows 10 SDK 1909 update

Microsoft has released an SDK update aligned with the Windows 10 November 2019 Update (version 1909), supporting UWP development. The latest WinUI 2.2 library, released in August, introduces a new TabView control.

Microsoft’s open‑source DAPR and OAM projects

DAPR (Distributed Application Runtime) is an open‑source runtime that simplifies building microservice applications across any language or framework, offering building blocks such as service invocation, state management, pub/sub messaging, and event‑driven bindings via gRPC or standard HTTP APIs. DAPR is currently in Alpha.

OAM (Open Application Model) defines a cloud‑native application specification that separates application logic from deployment and infrastructure details. It aims to be platform‑agnostic, supporting Kubernetes and other environments, and is being developed in collaboration with the Open Application Model Foundation.

Both projects are available on GitHub (https://github.com/dapr/dapr and https://github.com/oam-dev/spec/).